Four demonstration fields were selected to evaluate the effects of different irrigation management strategies in corn and soybean production. These locations were selected from southwest Michigan as shown in Figure 2 , since the majority of irrigated acreage is located within this region of the state (United States Department of Agriculture, National Agricultural Statistics Service, 2019). These areas possess the largest potential for improvement in water use efficiency using center pivot irrigation. Field locations, size, cropping regime, and soil type for each site are shown in Table 1 . In general, the area’s soil is well-drained and contains large proportions of sand.

Since this study was conducted in fields that were owned by private entities who were gracious enough to allow for manipulation of their irrigation systems, there were some factors which could not be held uniform across all fields. Tillage practices varied by farm and crop as seen in Table 2 .

Each field was divided into five plots to demonstrate irrigation prescriptions ( Figure 3 ). Table 3 describes the details on the threshold for Maximum Allowable Depletion (MAD) and irrigation volume for each irrigation treatment and each crop. Both the threshold for MAD and irrigation volume per application were determined based on irrigation simulations using Michigan State University Irrigation Scheduler Software. The irrigation volume for corn was larger than soybeans due to the rooting depth. MAD at 30%, 50%, and 60% were selected to evaluate the effects of MAD on the production and differentiate the total amount of irrigation application among the treatments. Fourth treatment was the producer’s typical irrigation management practice. The dry corner of each site was used as a non-irrigated treatment area for comparison of yields with irrigated areas. These dry corners of the field were outside the pivot’s area of coverage and were managed the same as the other treatment areas but had no irrigation applied.

Within each irrigation management zone, field conditions were monitored using the Low-Cost Sensor Monitoring System (LOCOMOS), which was developed by Michigan State University Irrigation Lab and was accompanied by at least three Soil Watch 10, manufactured by PinoTech (Zachodniopomorskie, Poland) (Holycross, 2023). The Soil Watch 10 sensor is a capacitance-based soil moisture sensor that operates at 75Mhz and has a probe length of 3.8 cm (Pinotech SoilWatch 10 Specifications, 2024). Sensor values were validated by comparing it with taking volumetric water content samples. The Soil Watch 10 calibration is described in Dong et al. (2024).

Sensors were installed at depths in and below the rooting zone to track water movement during and after each precipitation or irrigation event. Figure 4 displays how sensors were horizontally installed at a variety of depths in close proximity to the crop’s roots and not directly under the data logging system. For corn, sensors were placed at depths of 30.5, 61.0, and 91.4 cm. In soybeans, sensors were placed at depths of 15.2, 38.1, and 61.0 cm. LOCOMOS technology was utilized to measure precipitation using a rain gauge, manufactured by Davis Instruments (Hayward, CA, USA). Additionally, PHYTOS 31 leaf wetness sensors, manufactured by METER Group (Pullman, WA, USA) were integrated into the LOCOMOS to monitor the duration of leaf wetting, allowing for the potential correlation with foliar disease pressure. The sensor data were collected on an hourly basis and sent to a cloud website (locomos1.com) through an embedded cell modem.

In addition to sensor data, crop development was assessed on a weekly basis, especially during critical phases, enabling a more thorough analysis of yield impacts. Site scouting allowed for monitoring of any quantitative estimates of differences in disease presence. While related to yield, disease presence required quantification of impact for comparison of water use efficiency and yield across different irrigation management areas. Significant differences in disease presence were not found. Thus, data collected on disease presence was not included in this study and the effects of disease were assumed to be comparable between treatments. Furthermore, to compare crop performance, spatial yield data from the combine’s yield monitor was collected from each field at harvest. This data was then analyzed on an annual basis for each field, independently, to represent the efficiency of each irrigation management prescription. The average non-irrigated yield was subtracted from the average yield of each management area and divided by the volume of water applied to that particular management area to calculate water use efficiency, following Equation 1 (Irmak et al., 2011).

Where, Y_Irr represents the yield of an irrigated area (kg/ha), Y_Dry represents the yield of an non-irrigated area (kg/ha), and V_Irr is the volume of irrigation applied to the irrigated area (mm).

Statistics for this study were calculated using a combination of Microsoft Excel and R Studio. Microsoft Excel was utilized to transfer data over the various stages of this study and hold it for later refinement and analysis. Calculations of significance and power were reserved for R studio version 2023.12.0. Significance was analyzed using a pairwise Tukey’s test for significance at an alpha of 0.05. Once treatments were tested for significance, boxplots of yield and WUE were created to visualize the data.

Figure 5 shows weather summary during the growing season in Burlington/Union City, Parkville, and Sturgis, respectively. Data was collected from the National Weather Service starting in May and ending in August each year 2021 through 2023.

Comparisons between irrigated and non-irrigated yields are displayed in Figures 6 and 7 for all corn and soybean fields in this study. Corn yield data are summarized for all corn plots included in the study in Figure 6 , with a p-value of 0.115. Conversely, when comparing the averages of all irrigated corn areas to those of the non-irrigated corn areas, differences are much weaker than in the soybeans. Figure 7 shows a trend favoring irrigated soybeans that are statistically differentiable at an alpha of 0.1, due to a p-value of 0.0589.

Figure 7 shows the average corn yield and average WUE from all the demonstration fields over the duration of the study. Average corn yield was nearly identical for the under-, recommended, and grower’s treatment areas while the average of non-irrigated areas were lower than other treatments. WUE decreased as irrigation volume and frequency were increased in corn.

Soybean yield and WUE are summarized for all soybean plots included in the study. Figure 8 shows an average soybean yield and average WUE for 2021 to 2023. Yields were similar with the exception of the non-irrigated experimental treatment. WUE results follow a similar trend but include an increase in the upper range for the treatment 1 (60% MAD, under-irrigated) area and a decrease in the upper range for the treatment 3 (30% MAD, over-irrigated).

Statistical analysis of average yield for all treatments using Tukey’s pairwise comparison methods were conducted. All of the compared treatment statistics resulted in P-values above the value of alpha (0.05). Statistical analysis showed a p-value of 0.112 for non-irrigated and treatment 2 (50% MAD; recommended irrigation) and a p-value of 0.08 for non-irrigated and treatment 3 (30% MAD; over-irrigated in soybean production.

The relationship between water applications (irrigation and precipitation) on corn and soybean yields was observed. Figure 9 shows a slight upward trend in yield when irrigation and precipitation increase. Conversely, increased irrigation and precipitation in soybean production resulted in decreased soybean yields.

The economics of the applied irrigation treatments are calculated on each treatment’s three-year average for all corn and soybean fields in the study. Tables 4 and 5 shows economics of irrigating corn and soybeans by treatment, respectively. Calculations account for the average corn value in 2023 which was 0.0349 United States Dollars (USD) per kilogram, the average soybean value in 2023 of 0.0985 USD per kg, average yield of each treatment, and either diesel or electric pumping costs (United States Department of Agriculture, National Agricultural Statistics Service, 2024). These calculations assume two different rates for operating wells and center-pivot irrigation systems: an electric rate of 0.36 USD per hectare millimeter and a diesel rate of 1.029 USD per hectare millimeter (United States Department of Agriculture, National Agricultural Statistics Service, 2024). The columns labeled cost (diesel and electric) reflect the expense of pumping and powering the irrigation system per millimeter of irrigation applied to each hectare. Whereas the columns labeled profit (diesel and electric) display the value of the crop produced per millimeter of irrigation applied to each hectare.

The efficient management of water resources in agriculture is paramount for sustainable crop production and environmental stewardship. The impact of irrigation scheduling on WUE, defined as the improvement in yield divided by the amount of irrigation applied, was investigated in this study. WUE between the treatments from corn (P-value = 0.607) and soybean (P-value = 0.962) fields were not statistically significant different. In theory optimized irrigation scheduling practices, including the timing and amount of water application, have the potential to influence crop water use and performance. By aligning irrigation practices with crop water requirements throughout the growing season, growers should be able to enhance soil moisture availability, promote optimal plant growth, and mitigate the risk of water stress-induced yield losses. The results in the boxplots of Figures 7 and 8 do not display a statistically significant difference between the irrigation application treatments. The low number of samples and relatively high variability of field conditions are suspected to be the causes of insignificance. Therefore, the P-values are not low enough to prove a statistical difference at an alpha value of 0.05.

Although the volume of precipitation Michigan receives is occasionally sufficient in terms of maximizing crop yield, it is rare that the timing of precipitation events aligns with the crop’s demands. Hence the need for irrigation to meet the needs of the crop between precipitation events. However, over the course of this three-year study precipitation supplied enough water to limit the number of irrigation applications that could be made under the experimental design described in Table 3 . The relatively small volume applied to the treatment 2 (50% MAD; recommended irrigation) areas can be compared to the sum of each month’s precipitation. In years where more applications are required, there is a potential that results may differ from the trends observed in this study. In the case of the treatment 1 (60% MAD, under-irrigated), crop stress and evaporative losses incurred during the application of irrigation theoretically drive differences in yield and WUE. On the other hand, the treatment 3 (30% MAD, over-irrigated) experimental treatment should theoretically suffer in terms of water use efficiency due to the treatment’s increased total application volume and number of applications made. In both of these treatments, the critical factor driving a difference in water use efficiency is time under which precipitation is not sufficient, and irrigation may need to be applied. Every time a precipitation event fills the soil profile, there is a period in which all treatments are subject to the same conditions, reducing the opportunity for WUE to be impacted under this study’s experimental design. Additionally, the unusually high non-irrigated treatment area yields are also supportive of the idea that the impacts of the treatments were likely masked by frequent precipitation in the years of study. As seen in Figures 6 , the non-irrigated areas were unusually similar to the irrigated areas, with the corn fields not displaying a strong statistical difference (P-value=0.115). Total precipitation may have been similar to the thirty-year average precipitation for this region, it can also be seen that this volume comes in consistent intervals during some seasons and sporadically in others. In this study, fields included in the data set had between one and eight irrigation applications as permitted by the methods, creating a large difference in potential for effect to take place.

A more noticeable difference is displayed when the volume of water applied is accounted for in the WUE equation as compared to this study’s yields. While none of the treatments indicate a statistically significant difference in terms of WUE at an alpha of 0.05, it is noticeable that the averages for the different experimental treatments are not as similar as the averages for yield. It may be apparent that the total irrigation volumes applied per treatment have more impact on the resulting WUE than the yields do since yields were quite similar. However, the response observed is not a linear trend, even at reasonable irrigation volumes and thresholds. While frequent and heavy precipitation might be the cause of similar results for the treatment 3 (30% MAD, over-irrigated) and treatment 2 (50% MAD; recommended irrigation) areas, the treatment 1 (60% MAD, under-irrigated) should have been favored by the experimental design. While it’s possible that there was little opportunity for the impact of the treatments to take effect, it is also possible that different treatments may have been more appropriate at different times throughout the growing season. While application volumes were adjusted in the early stages of crop development to better suit the crops needs and rooting depth, it is likely that different experimental treatments may have aligned with certain precipitation events at specific times throughout the growing season. In the ideal scenario, the soil moisture would drop to the irrigation threshold value, then triggering irrigation, and then irrigation would be applied filling the soil profile to field capacity. After irrigation, the soil profile would dry back down to the irrigation threshold at the time of precipitation, completing a full cycle of the process. Conversely other experimental treatments which did not align with precipitation events would likely require an additional irrigation application or loose the impact of remaining soil moisture provided by the initial application that was pushed below the root zone by precipitation. This lends to the concept that a more flexible approach to irrigation scheduling may be required to observe its full potential.

The challenges posed by weather on irrigation management in sandy soils are significant and multifaceted, particularly in the case of more intense and irregular precipitation events. Sandy soils have unique characteristics that influence water retention and drainage, making them especially vulnerable to the impacts of changing weather patterns. One of the key issues with sandy soils is their low water-holding capacity. Unlike clay or loam soils, which can retain moisture for longer periods, sandy soils drain quickly, which can lead to water stress for crops, especially during dry spells. However, when faced with intense or irregular precipitation events, sandy soils can become saturated rapidly, leading to increased runoff and leaching. This creates a challenge for irrigation management, as there is limited capacity to store precipitation. Sandy soil’s relatively low water holding capacity is also a benefit as it increases drainage, reducing the risk of flooding which affords producers more control over their field’s soil moisture than a heavier soil. Maintaining soil moisture at unnecessarily high levels is inefficient due to the soil’s low soil water storage capacity for potential precipitation, which could lead to an increased risk of runoff if storage is rapidly exceeded. Managing this excess water effectively without stressing the crop, wasting resources, and leaching valuable nutrients is a challenge, especially in humid climate regions. In regions experiencing more intense and irregular precipitation events, irrigation strategies must adapt to accommodate fluctuations while maintaining efficiency and practicality. This highlights the importance of selecting an irrigation volume and threshold that accounts for the soil characteristics, crop, and weather conditions. Additionally, these results lead to the implication that a more pliable irrigation scheduling method would be ideal.

Efficient irrigation management requires striking a delicate balance between utilizing precipitation and supplementing it with irrigation when necessary. This balance becomes increasingly difficult to achieve with erratic weather patterns. As seen in Figure 5 , the area of this study has received an increasingly inconsistent amount of precipitation in recent years. The irrigation needed to improve yields and water use efficiency is typically infrequent and relatively small in volume as compared to the total demand for corn and soybeans grown in Michigan. This underscores the need for precise timing of each irrigation application and calculation of irrigation volume based on soil, crop, rooting depth, and forecasted precipitation.

Weather forecasting, especially the prediction of precipitation timing and volume, has been a historical challenge. It has been found that the reproductive stages of corn and soybeans are more sensitive to fluctuations in soil moisture, thus, this period contains the highest potential for irrigation to impact yield (Payero, 2009). While this study focused on adhering to the experimental design during the same critical stages, significant differences in WUE were not observed. This is likely due to the availability of soil moisture due to precipitation during critical stages and the smaller differences between experimental treatments utilized in this study. In Figure 9 , corn and soybeans reacted differently to increasing combined volumes of irrigation and precipitation. Corn has a positive trendline indicating that increased irrigation and precipitation can increase yield under the right conditions in this area. This suggests that, for corn, additional water through either precipitation or irrigation can contribute positively to productivity, especially during periods of limited natural precipitation. In contrast, the relationship between water input and soybean yields was less straightforward. While some increase in water availability is generally beneficial, the data showed that excessive irrigation and precipitation were associated with reduced soybean yields. This counterintuitive trend may be attributed to waterlogging, overly saturated soil conditions, as well as frequent wetting, which can impair root development, promote plant diseases, and ultimately compromise overall plant health. Although disease presence was minimal or consistent between treatments in this study, it is possible that disease played a role that was not observed. For instance, a study on white mold suggests that soybeans have a disease severity index that increases with increasing total precipitation, when total precipitation in the month of July is between 20 mm and 108 mm (Fall et al., 2018). Overall, these findings underscore the importance of carefully managing irrigation strategies.

In another study where frequent and substantial precipitation took place during the growing season, statistically significant differences were observed between experimental treatments, allegedly due to a more pliable scheduling technique with more sophisticated weather forecasting (Mahdavi and Fujimaki, 2024). As described by methods of supplemental irrigation, efficient irrigation methods in the particular location of this study apply a volume just large enough to maintain soil moisture above the irrigation threshold until the next precipitation event. This may indicate that directly observing available water through the use of in situ soil moisture monitoring could be more advantageous than calculating the current available water as done in ET-based scheduling methods. Utilizing irrigation scheduling with the aid of soil moisture monitoring reduces potential errors in estimating current conditions while maintaining the best approach to forecasting the future conditions of the soil profile and crop’s needs. In a study with more strenuous water constraints, it was found that irrigation scheduling methods that combined multiple techniques were over 50% more effective than fixed interval and frequency applications (De Pascale et al., 2011). This study’s results, although not statistically differentiated, suggest the ideal volume is somewhere between applying just enough to maintain the soil moisture above the irrigation threshold until the time of the next precipitation event and applying a volume that does not exceed the water holding capacity of the profile at the time of the next precipitation event. Similar approaches to improving irrigation application timing and volume have been attempted using mathematical and observational methods for several decades with success (Stegman, 1983). How close one allows the soil moisture to get to the irrigation threshold and field capacity depends on one’s perceived values of water and crops as resources. It is likely that implementing irrigation scheduling methods and utilizing advanced monitoring technologies in conjunction has the potential to mitigate the impacts of weather variability on crop production and improve water use efficiency in sandy soil environments. Comparing the WUE of the grower’s methods in corn ( Figure 7 ) to any of the other experimental treatments shows the potential of irrigation scheduling methods.

As with any practice, the costs of irrigation must be outweighed by the benefits in order to justify making the investment in an irrigation application. One of the biggest obstacles to the adoption of irrigation management practices is a lack of cost effectiveness, even when water use efficiency is high (Koech and Langat, 2018). Tables 4 and 5 show that although the yields favor higher irrigation volume treatments, the average profits for both corn and soybeans support the treatment 2 (50% MAD; recommended irrigation) and treatment 1 (60% MAD, under-irrigated) treatment. In Michigan, 2,306 irrigation systems and wells are still powered by diesel (United States Department of Agriculture, National Agricultural Statistics Service, 2024). The price of operating diesel powered pumps and irrigation systems has increased dramatically in the last decade, for the calculations in this study diesel pumping cost is assumed at about 10 USD per hectare cm. Considering the recent increase in diesel prices, this is a conservative estimate that drives the economic choice to favor applications similar to the treatment 2 (50% MAD; recommended irrigation) or under irrigated treatment, even more so than the economics of electric power. Labor costs were not included in this calculation as all center pivots in this study were operated remotely. Additionally, the labor costs of center pivot irrigation per acre are generally low in comparison to improvements in profit. Likewise, the fixed costs of the center pivot, pump, and accompanying hardware were not included as they are assumed to be constant values across the different treatments. Even while only assessing the impact of pumping the water for irrigation and powering the center pivot itself, a calculation of the profit values from Tables 4 and 5 , applied to a sizable field reveals the potential for thousands of dollars of impact between many of the treatments. Although economics are essential to making cost effective irrigation applications, they are far from the only value one should consider. In this study, significant differences in yield and WUE were not observed, thus there is little justification to apply more than what the treatment 2 (50% MAD; recommended irrigation) scheduling suggests. However, there is reasoning to apply more than the treatment 1 (60% MAD, under-irrigated) scheduling suggests, based on the economics and mitigation of risk to plant health. It is logical to buffer the system by applying a slightly larger volume of irrigation and maintain the soil moisture above the crops wilting point due to the unpredictability of climatic conditions, so long as the selected volume does not exceed the soil’s water holding capacity. There were multiple instances in this study where the peak demand of the crop landed on a dry spell, this caused the available water in the under irrigated area to quickly diminish due to the application volume not filling the profile. This led to far more frequent applications, which in a dry growing season could become costly and induce unnecessary stress on the crop.

This study covers three growing seasons and provides important insights into irrigation management under humid climate conditions, but several limitations should be acknowledged. The research was conducted within a limited geographic region in southwest Michigan, which may constrain the applicability of the findings to other locations with different environmental or agronomic conditions. Although data spanning multiple seasons enhance the temporal relevance of the results, variability in annual weather patterns, soil properties, and other uncontrolled environmental factors may have influenced treatment effectiveness and yield responses. Additionally, the number of replication sites and treatments was limited, reducing the statistical power to detect subtle but potentially meaningful differences among irrigation strategies. Further research with expanded geographic scope, increased replication, and controlled variability is needed to refine these findings and support broader generalization.