A VBA-Based Field Water Balance Model for Efficient Irrigation Water Management of Corn (Zea mays L.)

Abstract

A field water balance model for efficient irrigation water management of corn was developed using Excel VBA. The model consists of five sub-components or modules, namely, (1) a plant subcomponent, (2) an effective rainfall subcomponent, (3) an evapotranspiration subcomponent, (4) a soil water dynamics subcomponent for the modeling of water flow into and within the soil layers, and (5) an irrigation subcomponent for the estimation of the required amount and timing of irrigation. The model was calibrated and validated using observed data from field experiments and the results showed a reasonably good agreement between the observed and simulated soil moisture values (MAE = 5.76 mm to 12.00 mm, RMSE = 6.83 mm to 13.12 mm, NRMSE = 0.102 to 0.196, and NSE = 0.37 to 0.90). The simulations emphasized that a significant amount of water savings can be achieved when rainfall is properly accounted for in managing water in the field, and that the frequency of rainfall occurrences is as important as the magnitude of rainfall received by the crops. The wide-ranging user-friendliness and simplicity of the model developed in this study can pave the way to eliminating the barriers which cause farmers to resist advancements in their farming practices as the model can easily be used not only by researchers and scientists but also by farmers, especially those with basic knowledge of spreadsheets.

1. Introduction

Corn ranks as the second most important crop in the Philippines. This commodity is used not only as a substitute staple food, especially during seasons of rice shortage, but is also a primary source of feed in the animal industry of the country [1]. A total of 1.46 million farm households depend on the corn industry for their livelihood. In terms of crop share to the gross value added (GVA) in the agricultural crop subsector, corn also ranks second at 5.7% [2].

The Philippines had an annual total corn production of more than 4 million metric tons since 1987, except in 1998. After the country recorded a peak production of 4.85 million metric tons in 1990, corn production in the succeeding years decreased at an average annual rate of 2.8% until the lowest production of 3.82 million metric tons was recorded in 1998. The industry immediately recovered in 1999, when a total corn production of 4.58 million metric tons was recorded, or a 19.9% increase in production was observed. Over the next 21 years (2000–2020), corn production in the Philippines increased at an average annual rate of 2.2%. In 2020, the corn sector of the country recorded a total production of 8.12 million metric tons, an all-time high record in the country which is 1.75% higher than the previous year’s production of 7.98 million metric tons [3].

In terms of the area harvested for corn, a similar trend to that of corn production was observed from 1990 to 1998. The total area harvested for corn was also the highest in 1990 at 3.82 million hectares. In the succeeding years, however, the total area harvested for corn declined at an average annual rate of 5.76%, recording the lowest in 1998 at 2.35 million hectares. The area harvested for corn increased by 12% the following year. However, an average decline of 0.11% was observed in the area harvested for corn over the next 21 years (2000–2020) [3].

Although the production of corn has continued to increase over the years despite the decrease in the area allocated to corn, the rate of increase in the volume of corn produced is still significantly lower than the more rapid increase in demand. Several factors contribute to the decreased corn production and declining corn farmland in the Philippines. These factors include poor soil fertility, harm from pests and diseases, insufficient farmer income, unpredictable weather patterns and extremes, limited access to irrigation, and lack of knowledge on irrigation scheduling [1]. This study focused on addressing the latter by developing a decision tool that can guide corn farmers in providing the correct volume of irrigation to the crop at the right time.

To optimize irrigation timing for corn production, a good understanding of how corn water needs fluctuate during its entire growing season is necessary. One of the characteristics of corn that is highly important in irrigation water management is its ability to thrive despite a lack of rain or water shortages during its early, vegetative, and after-ear-formation stages. To enhance corn yield and grain quality, adequate water should be made available to the plants throughout the growing period in the reproductive stage [4]. Hence, a carefully planned irrigation schedule guided by the growth process of the plant is of utmost importance in managing water for corn production, especially when water is a limiting factor.

In general, irrigation water application in the upland cropping system of the Philippines is based on either a predetermined schedule or the farmer’s physical inspection of the soil and the crop in the field. Irrigation events are seldom based on crop water needs. Instead, farmers employ a fixed irrigation cycle at a certain time interval which is attributed to their lack of knowledge on the theoretical basis of irrigation scheduling [5]. The unfamiliarity of farmers with producing upland crops under irrigated conditions is prevalent in the Philippines. Such unfamiliarity with means of enhancing irrigation application exists among corn farmers in the country [6].

Another problem with the current corn production systems is the misconception on the part of farmers that corn does not need irrigation, which is one of the reasons why corn production systems in the Philippines are mostly rainfed. In areas where there are irrigation services, corn farmers still rely on seasonal rains and request irrigation only when their farms are subjected to prolonged dry scenarios [6]. Aside from the very high risk associated with the nonoccurrence of expected rains on which corn production greatly depends, such water management practice can lead to either overirrigation, where too much irrigation water is applied than what is needed by the crops at noncritical stages, or underirrigation where an insufficient amount of water is applied to the crops during critical stages which can lead to water stress and affect the yield. These scenarios contribute to the relatively low water productivity observed in crop production.

In view of the aforementioned problems, new water management techniques that will reduce water use in agriculture while maintaining high crop yields should be developed. This need is not only important but also urgent considering that the Philippine government is planning to expand the irrigation systems and facilities for diversified crops such as corn [7]. One of the most convenient approaches to efficiently managing irrigation water for upland crops such as corn is through the use of water balance simulation models.

In recent years, the contributions of simulation models in the scrutiny of the vital processes occurring between the soil, water, atmosphere, and plants have increased significantly [5]. Several computer-aided simulation models anchored on the concept of water budgeting over the root zone in estimating crop water requirements have already been developed [8,9,10,11,12]. These models have high acceptance among researchers and other professionals in the field of irrigation but not local farmers. Farmers’ acceptance of simulation models for irrigation scheduling has been gradual due to two major reasons: (1) the models’ interface is not user-friendly and operating the models requires huge computers which are inaccessible to most farmers, and (2) lack of appropriate local values to represent reference crop evapotranspiration (ETo). Due to this challenging situation, more accessible platforms and farmer-friendly interfaces of irrigation scheduling simulation models should be developed [13].

To date, no study exists in published peer-reviewed literature on the development of a farmer-friendly water balance model that can be used for irrigation water management of corn. Hence, this study aimed to develop a water balance model and an irrigation scheduler that is easy to use and operate, using a more accessible platform that does not require huge computers and a model that requires commonly available input data. This study is the first attempt to develop a tool to improve the water use efficiency of corn production systems in the Philippines through the development of a simple spreadsheet-based simulation model.

2. Materials and Methods

2.1. Site Description, Characterization, and Data Acquisition

The field experiment was conducted at the Central Experiment Station (CES) of the University of the Philippines Los Baños (UPLB), Laguna. Experimental plots, arranged in a complete randomized design with three replications, were established in Area B7 (14°9′54″ N and 121°15′9″ E). Additional pot experiments were performed at the Institute of Agricultural and Biosystems Engineering (IABE) Complex, UPLB, which was only 200 m away from Area B7. The experimental setups were adjacent to the National Agrometeorology Station (NAS) where the 30-year historical climatic records were obtained.

Two soil sampling techniques were employed to facilitate spatial and depth variability analysis of various soil properties of the site. An unaligned grid sampling technique was employed to collect samples that were used in determining the spatial variability of the soil texture in the experimental area. Using the vector grid tool in QGIS 2.18.10, 25 m by 25 m grids enclosing the entire field were created. From each grid, sampling points were determined from two points located randomly inside the grid. Soil samples with a depth of 0–25 cm from the surface were collected. Random sampling was performed to collect soil samples from the other layers of the soil profile. Five samples were taken randomly from the site. For each sampling point, soil samples with a depth of 25–50 cm, 50–75 cm, and 75–100 cm measured from the surface were collected to determine soil characteristics such as texture, bulk and particle density, and porosity.

Laboratory analysis was also performed to obtain information on the capacity of the soil in the site to hold water. The water retention properties of the soil in the field were determined in the pF Laboratory at the International Rice Research Institute using pressure plate apparatus. A summary of the different soil properties determined in the study area is shown in

Table 1. Soil properties determined in the study area.

Layer	Depth (m)	Moisture Content at Field Capacity (cm3 cm−3)	Moisture Content at Permanent Wilting Point (cm3 cm−3)	Texture	Sand (%)	Clay (%)
1	0.00–0.30	0.323	0.157	Clay Loam	36.6	37.0
2	0.30–0.45	0.321	0.155	Clay Loam	37.0	40.1
3	0.45–0.60	0.325	0.160	Sandy Clay Loam	55.7	27.0
4	0.60–1.00	0.330	0.153	Clay Loam	41.6	39.4

.

Sweet Grande F1, a hybrid sweet corn variety bred by East West Seed Inc., was used in this study. This corn variety is high yielding, has vigorous growth, and is characterized by big ears with sweet and tender kernels. Sweet Grande F1 is well-suited for both wet and dry season planting and has a growth duration of 75–78 days after seeding. For the program, default values of some basic crop parameters required in the simulation were generated in this study. Crop characteristics determined experimentally included growth stage-specific properties such as rooting depth and maximum allowable depletion (MAD) for high crop yield.

To calibrate and validate the model, pot and field experiments were set up for soil moisture monitoring. Experimental pots filled with loose soil from CES were prepared to simulate soil conditions in the field and were laid out in a randomized complete block design to account for the effects of spatial variation in the area. Soil moisture content measurements were taken throughout the initial stage of the crop using a FieldScout TDR 300 soil moisture meter. The measurements obtained from the pot experiments were used to calibrate the model. Similarly, field experiments were conducted to validate the accuracy of the model’s simulation. Four 6 m × 10 m plots were established at our project site at CES Area B7. Scheduling of irrigation in the three plots was based on a non-uniform water application depending on the depth of the root zone and MAD/critical moisture level for corn at different stages of growth (Refer to

Table 2. Maximum allowable depletion percentages used in the simulation.

Growth Stage	Maximum Allowable Depletion (%)	Effective Rooting Depth (mm)
Initial	55	300
Crop Development	50	450
Mid-season	50	600
Late-season	60	600

Source: Painagan and Ella, 2022 [14]; Allen et al., 1998 [15]; USDA, 1997 [16].

). Soil moisture content measurements at 15 cm, 22.5 cm, 37.5 cm, and 52.5 cm depth were taken throughout the growing stage of the crop using the FieldScout TDR 300 soil moisture meter. Three measurements for each of the specified depths were taken around the drip emitters and the average of the measurements from each soil depth was treated as the moisture content of the field at that depth. The measurements obtained from the field experiments were used to validate the model.

2.2. Development of the Model

2.2.1. Model Description and Conceptual Framework

The spreadsheet-based model was based on the concept of water balance as a simple statement of the conservation of mass. This means that the amount of water stored in a given soil volume at a specific time interval must equal the difference between the volume of water added to the system and the volume of water lost by the system during that specific time interval. In this study, water balance modeling was performed at the effective rooting depth of the plants and was simulated at various desired time intervals (Figure 1.). The working equation was expressed as:

$$S{W}_{t }=S{W}_{t-1 }+P_{et }-R_{t }+I_{t }–ET{a}_{t }-D{P}_{t }+C{R}_{t }$$

(1)

where SW_t is the soil water stored in the effective root zone on time t (mm), SW_t−1 is the soil water stored on the previous time (mm), P_et is the effective rainfall on time t (mm), R_t is the runoff from the soil surface on time t (mm), I_t is the irrigation water infiltrating the soil on time t (mm), ET_at is the actual crop evapotranspiration on time t (mm), DP_t is the deep percolation on time t (mm,) and CR_t is the capillary rise from shallow water table on time t (mm).

The general workflow of the spreadsheet-based model is presented in Figure 1. The key function of the general workflow is to initialize the simulation procedures by preparing the model parameters required in either the main program or in the subcomponents. As soon as the present soil moisture condition is entered, the main program will either call the irrigation subcomponent to compute the amount of irrigation if the initial condition is below the threshold set or call the other subcomponents to model the depletion in the soil and determine the next irrigation event based on the threshold established.

The program has five sub-components or modules: (1) plant subcomponent for the input of the required plant characteristics; (2) effective rainfall subcomponent for the estimation of the portion of rainfall that is useful or utilizable for crop production; (3) evapotranspiration subcomponent for the estimation of the actual evapotranspiration of the crop over the growing period; (4) soil water dynamics (infiltration and redistribution) subcomponent for the modeling of water flow into and within the soil layers; and 5) irrigation subcomponent for the estimation of the required amount and timing of irrigation. Aside from these subcomponents, sheets for a graphical presentation of the simulation and notable relationships, and a sheet for instructions for the simulation were incorporated into the spreadsheet-based program.

In conceptualizing the model, the use of a crop water requirement to provide enough water based on its crop water demand was emphasized. The program operates within the context that crop water needs fluctuate during its entire growing season. In this study, an irrigation schedule was generated based on the four growth stages of corn. The crop’s ability to thrive despite the lack of rain or water shortages during its early, vegetative, and after-ear-formation stages and its high demand for water during its reproductive stage was accounted for by varying the maximum allowable depletion used to trigger irrigation in each growth stage. The program also considered varying the effective rooting depth of the crop to provide an added avenue for saving water. The effective rooting depth used in the model was based on the crop growth stage since the crop has shallower access to soil moisture due to shallow rooting depth during early growth stages. Consequently, a relatively lower volume of water is also needed to fill the rooting zone of the crop with water whenever an irrigation event is triggered during these stages. The maximum allowable depletion and the effective rooting depth used in this study are shown in Table 2.

The program estimates the effective rainfall in a module that requires daily input of rainfall depth in millimeters. Upon clicking the simulate button, the program, by default, will simulate moisture level fluctuation in the soil without accounting for rainfall. In case a rainfall event occurs at any date during the growing period of the crop, the user can enter the amount of rainfall that has occurred and activate the simulate button again to update the simulation, thereby accounting for the entered rainfall depth. The program lacks the capacity to provide estimates of dependable rainfall. Hence, a user’s good knowledge of the distribution of rainfall to properly schedule farming activities is an advantage in using the program. Moreover, at least an improvised rain gauge is required if the field is away from a weather station.

In estimating crop water requirements, the evapotranspiration module operates only with a historical record of climatic data. It is not equipped with codes that would allow real-time calculation of evapotranspiration when daily climatic data are accessible to the farmers. It does not allow computation of potential evapotranspiration using other methods either. Hence, to use the program, a historical climatological record from a nearby weather station is required. In this study, a 30-year record of climatic data obtained from the NAS-UPLB was used in the simulation. The climatic data for the computation of reference crop evapotranspiration using the FAO Penman–Monteith equation [15] include maximum and minimum temperatures, wind speed, relative humidity, and solar radiation. In this study, the crop coefficients used were obtained from the recommendations of Allen et al., 1998 [15] and USDA, 1997 [16].

Once the effective rainfall and evapotranspiration are estimated, the soil water dynamics module will require at least the initial moisture content of the effective rooting depth of the crop and the soil type at this depth to proceed with the simulation of soil moisture depth. The average moisture content of the soil once the crop has emerged and the soil type at the required depths should be provided by the user. If the user has the capacity to determine soil properties in the laboratory, the following data can be entered to obtain a more accurate simulation: field capacity on a volume basis, the permanent wilting point on a volume basis, sand percentage, and clay percentage. Otherwise, the program will use average values of these properties based on the inputted soil type. The average values of soil properties stored in the program were derived from published literature [17]. Once the input parameters needed were complete, the simulation was initialized, and water balance was performed in layered soil where the water flow in the soil layers is assumed to occur from the center of the first soil layer to the center of the second layer until it reaches the last layer [18]. Other soil properties vital in the water balance approach were estimated using different widely used equations discussed earlier. The conceptual framework of the developed program is presented in Figure 2.

2.2.2. Water Balance Model Subcomponents, Data Requirements, Interface, and Database

Plant Subcomponent

The plant subcomponent allows the user to describe the characteristics of the crop to be used in the simulation. Crop characterization will aid the model in establishing crop parameters required in simulating processes either in the main program or in other subcomponents. In this subcomponent, the user has the option to provide the necessary input data or use default values the model provides, which are either results of experiments conducted in this study or accepted values taken from published literature [15,18,19]. Crop characteristics provided with default values can be altered by the user, particularly the following: crop name and variety, duration (in days) of the four major crop growth stages (initial, development, mid-season, and late-season), expected rooting depth, root extraction percentage, and depletion fraction threshold or maximum allowable depletion (MAD) for high crop yield at the stages identified.

The plant subcomponent is equipped with a tool that calculates the length of the duration of the four major crop growth stages in the program. The tool estimates the length of each growth stage as a percentage of the total crop duration from emergence up to maturity or harvest. The initial stage is estimated as 20% of the total crop duration, while the crop development is estimated immediately after the initial stage as the next 30% of the whole crop days to maturity. The mid-season stage is estimated immediately after the crop development stage as the next 25% of the total crop duration, while the late-season stage is estimated immediately after the mid-season stage as the next 25% of the whole crop days to maturity.

The plant subcomponent also allows varying the rooting depth of the crop at a specific growth stage. The user may enter the effective rooting depth at each stage or use the default values determined from an experiment conducted using the corn variety chosen for this research. The plant subcomponent is also equipped with a graphical tool that plots the day at which the crop emerges until the time it reaches maturity. This will help the user in identifying crop management needs at different growth stages.

Evapotranspiration Subcomponent

The evapotranspiration subcomponent estimates water losses by two distinct but simultaneous processes, namely soil surface evaporation and crop transpiration. Reference evapotranspiration was computed using FAO Penman–Monteith equation (Equation (2)), and crop coefficient values were obtained from FAO Irrigation and Drainage Paper No. 56 [15].

$$E{T}_o=\frac{\mathbf{0.408}∆\left(R_n-G\right)+\gamma \frac{\mathbf{900}}{T+\mathbf{273}}{u}_{\mathbf{2}}\left(e_s-e_a\right)}{∆+\gamma \left(\mathbf{1}+\mathbf{0.34}{u}_{\mathbf{2}}\right)}$$

(2)

where ET_o is the reference evapotranspiration (mm day⁻¹), R_n is the net radiation at the crop surface (MJ m⁻² day⁻¹), G is the soil heat flux density (MJ m⁻² day⁻¹), T is the mean daily air temperature at 2 m height (°C), u₂ is the wind speed at 2 m height (m s⁻¹), e_s is the saturation vapor pressure (kPa), e_a is the actual vapor pressure (kPa), e_s − e_a is the saturation vapor pressure deficit (kPa), Δ is the slope vapor pressure curve (kPa °C⁻¹), and γ is the psychrometric constant (kPa °C⁻¹).

This subcomponent is linked to the plant subcomponent for retrieving crop phenology data, such as the length of the four distinct stages of the crop, the planting date, the date upon emergence, and the date of harvest, among others. In this subcomponent, the differences in the water demand for evapotranspiration during various growth stages of the crop are considered in the estimation. The differences were accounted for by the daily estimation of Kc of the crop throughout the growing period. For each growth stage, the corresponding crop coefficient (Kc) was selected from the recommended values of crop coefficient from FAO Irrigation and Drainage Paper No. 56 [15]. The actual crop evapotranspiration value used in the main program was calculated as the product of the reference crop evapotranspiration and crop coefficient (Equation (3)).

$$E{T}_c=E{T}_o \times K_c$$

(3)

where ET_c is actual crop evapotranspiration (mm day⁻¹), ET_o is the reference evapotranspiration (mm day⁻¹), and Kc is the crop coefficient.

This subcomponent allows the partitioning of actual evapotranspiration into four soil layers used in the simulation by using the characteristic extraction pattern for most crops or the 4-3-2-1 rule: 40% of extracted water comes from the top 25% of the root zone, 30% comes from the second 25%, and so on. The subcomponent also utilizes historical records of climatic data. The user is required to provide at least 30 years of daily climatic records in a simple database in one of the worksheets in the workbook. Reference evapotranspiration was generated based on the 80% probability of occurrence of the weather parameters required, and daily values of Kc were derived from the curve created from recommended values of crop coefficient based on FAO Irrigation and Drainage Paper No. 56 [15].

Effective Rainfall Subcomponent

In the effective rainfall subcomponent, users should provide daily rainfall data as input. At least an improvised rain gauge is required as instrumentation and should be set up at the center of the field if there is no nearby weather station. Any rainfall occurrence during the cropping period should be reported in the model to allow real-time irrigation water management. Though this option is designed to perform real-time monitoring of water status in the field, the program will still provide a simulation of the next irrigation event using two different scenarios: (1) assuming no rainfall occurrence, and (2) accounting for the amount of rainfall that has occurred based on a weather monitoring station.

When rainfall data has been inputted, the effective rainfall subcomponent will then compute the portion of the reported rainfall that will be used in the simulation known as the effective rainfall. Effective rainfall can be defined as the portion of rainfall that is useful or utilizable for crop production. In view of this concept, precise knowledge about effective rainfall within the next 24–48 h is required for planning the next irrigation application where operation schedules of the laterals, for instance, may subsequently be adjusted. This will result in a better utilization of available water resources and improved crop production. In this study, effective rainfall was computed using the formulation developed for (a) shallow-rooted crops, and (b) medium/deep-rooted crops under a humid or sub-humid environment [20]. The developed method is given as follows:

(a)

Shallow-rooted crops

• If P_t < 3 mm, P_et = 0
• If 3 mm < R_t < (ET_a + RT_d), P_et = C × P_t
• If P_t ≥ (ET_a +RT_d), P_et = ET_a + RT_d = ET_a + (SMC_p − SM_i − I_t)

(b)

Medium-rooted crops

• If P_t < 5 mm, P_et = 0
• If 5 mm < P_t < (ET_a + RT_d), P_et = C × P_t
• If P_t ≥ (ET_a +RT_d), P_et = ET_a + RT_d = ET_a + (SMC_p − SM_i − I_t)

where P_t is the total rainfall for the considered time period (mm), P_et is the effective rainfall (mm), Et_a is the actual evapotranspiration (mm), RT_d is the crop root zone deficit computed as SMC_p − SM_i − I_t (mm), SMC_p is the soil moisture storage capacity (or FC) of the root zone soil (mm), SM_i is the initial (at the start of the considered period) soil moisture content (mm), I_t is the amount of irrigation (if applied within the considered period) (mm), and C is the coefficient representing the percentage term (~0.8–0.9).

Soil Water Dynamics Subcomponent

This subcomponent models water flow based on the assumption that water flow occurs from the center of the first soil layer to the center of the second layer until it reaches the last layer. In this program, the entire soil profile that covers the effective rooting depth of the crop was divided into 4 soil layers. Water flow in these layers is described by Darcy’s Law in which the net water flux, qnet (m/day), into soil layer i is expressed as the difference between the water flux entering and leaving the soil layer i, ${\mathrm{q}}_{\mathrm{i}}-1$ and ${\mathrm{q}}_{\mathrm{i}}+1$, respectively. A positive change in water flux signifies an increase in the water content of the soil while a negative change in water flux means that the moisture content is decreased. In the case of this program, the soil properties related to water flux were estimated using established and widely used equations.

The volumetric soil moisture content at saturation, field capacity (FC), and permanent wilting point (PWP) of each layer of the soil were computed using the equations proposed by Saxton et al. [21]. The proposed equations satisfactorily provided general estimates of soil moisture characteristics using more readily available data such as soil texture. Moreover, the equations allow easier calibration whenever particle size distribution data and field data on soil moisture characteristics are available. The FC, PWP, and saturated moisture content on a volume basis were estimated using the following equations:

$${\theta }_{FC,i}=exp\left[\left(\mathbf{3.4965}-lnA\right)/B\right]$$

(4)

$${\theta }_{PWP, i}=exp\left[\left(\mathbf{7.3132}-lnA\right)/B\right]$$

(5)

$${\theta }_{sat,i}=exp\left[-\mathbf{4.396}-\mathbf{0.0715}\left(\%Cl\right)-\mathbf{4.880}\times {\mathbf{10}}^{-\mathbf{4}}{\left(\%S\right)}^{\mathbf{2}}-\mathbf{4.285}\times {\mathbf{10}}^{-\mathbf{5}}{\left(\%S\right)}^{\mathbf{2}}\left(\%Cl\right)\right]\mathbf{100}.$$

(6)

$$A=exp\left[-\mathbf{4.396}-\mathbf{0.0715}\left(\%Cl\right)-\mathbf{4.880}\times {\mathbf{10}}^{-\mathbf{4}}{\left(\%S\right)}^{\mathbf{2}}-\mathbf{4.285}\times {\mathbf{10}}^{-\mathbf{5}}{\left(\%S\right)}^{\mathbf{2}}\left(\%Cl\right)\right]\mathbf{100}$$

(7)

$$B=-\mathbf{3.140}-\mathbf{2.22}\times {\mathbf{10}}^{-\mathbf{3}}{\left(\%Cl\right)}^{\mathbf{2}}-\mathbf{3.484}\times {\mathbf{10}}^{-\mathbf{5}}{\left(\%S\right)}^{\mathbf{2}}\left(\%Cl\right)$$

(8)

where for each soil layer i, ${\mathsf{\theta }}_{\mathrm{FC}}$ is the volumetric soil moisture content at saturation (m³/m³), ${\mathsf{\theta }}_{\mathrm{PWP}}$ is the volumetric soil moisture content at saturation (m³/m³), ${\mathsf{\theta }}_{\mathrm{sat}}$ is the volumetric soil moisture content at saturation (m³/m³), %Cl is the clay content of the soil (%), and %S is the sand content of the soil (%).

Soil hydraulic conductivity was estimated using the method proposed by Saxton and Rawls [22]. The saturated hydraulic conductivity was estimated using Equation (9),

$$K_{s, i}={\left({\theta }_{s,i}-{\theta }_{FC,i}\right)}^{\mathbf{3}-{\lambda }_i}$$

(9)

$${\lambda }_{ i}=\frac{\mathbf{1}}{B_i}$$

(10)

$$B_{ i}=\frac{ln\mathbf{1500}-ln\mathbf{33}}{ln{\theta }_{FC,i}-ln{\theta }_{PWP,i}}$$

(11)

where for soil layer i, K_s is the saturated hydraulic conductivity (m day⁻¹), θ_s is the saturated volumetric water content (m³ m⁻³), θ_FC is the volumetric water content at field capacity (m³ m⁻³), θ_PWP is the volumetric water content at permanent wilting point (m³ m⁻³), and λ_i is the slope of the logarithmic suction–soil moisture curve.

The unsaturated hydraulic conductivity was estimated using Equation (12),

$$K_{\theta , i}=K_{s, i}{\left(\frac{{\theta }_i}{{\theta }_{s,i}}\right)}^{\mathbf{3}+\mathbf{2}/{\lambda }_i}$$

(12)

where for soil layer i, K_θ is the unsaturated hydraulic conductivity (m day⁻¹), θ_i is the initial volumetric water content (m³ m⁻³), θ_s is the saturated volumetric water content (m³ m⁻³), and $\mathsf{\lambda }$ is the slope of the logarithmic suction–soil moisture curve.

Water flux was calculated using Darcy’s Law as expressed in Equation (13),

$$q={\overline{K}}_{\theta ,i} \frac{\left(H_{t,i-\mathbf{1}}-H_{t,i}\right)}{z_{i-\mathbf{1}}-z_i}$$

(13)

where for soil layer i, q is the unsaturated hydraulic conductivity (m day⁻¹), ${\overline{\mathrm{K}}}_{\mathsf{\theta },\mathrm{i}}$ is the logarithmic mean of hydraulic conductivities of layer i and i − 1 (m day⁻¹), z is the thickness of the soil layer (m), and H_t is the total hydraulic head (m).

The logarithmic mean of hydraulic conductivities was computed using Equation (14),

$${\overline{K}}_{\theta ,i}=\frac{K_{\theta ,i-\mathbf{1}}-K_{\theta ,i}}{ln\left(K_{\theta ,i-\mathbf{1}}\right)-ln\left(K_{\theta ,i}\right)}$$

(14)

where for soil layer i, ${\mathrm{K}¯}_{\mathsf{\theta },\mathrm{i}}$ is the logarithmic mean of hydraulic conductivities of layer i and i − 1 (m day⁻¹), ${\mathrm{K}}_{\mathsf{\theta },\mathrm{i}}$ is the hydraulic conductivity of layer i (m day⁻¹), and ${\mathrm{K}}_{\mathsf{\theta },\mathrm{i}-1}$ is the hydraulic conductivity of layer i − 1 (m day⁻¹).

The total hydraulic head was computed using Equation (15),

$$H_{t,i}=H_{m,i}+H_{g,i}$$

(15)

where for soil layer i, ${\mathrm{H}}_{\mathrm{t}}$ is the total hydraulic head (m), ${\mathrm{H}}_{\mathrm{m}}$ is the soil matric suction (m), and ${\mathrm{H}}_{\mathrm{g}}$ is the gravity head (m).

The gravity head was computed using Equation (16),

$$H_{g,i}=z_i$$

(16)

where for soil layer i, ${\mathrm{H}}_{\mathrm{g}}$ is the gravity head (m) and z is the thickness of the soil layer (m).

The soil matric suction was computed using Equation (17), and the needed parameters A_i, ${\mathsf{\Psi }}_{\mathrm{e},\mathrm{i}}$, and ${\mathsf{\Psi }}_{\mathrm{et},\mathrm{i}}$ were computed using Equations (18)–(20).

$$H_{m,i}=\{\begin{array}{ll}\mathbf{3.3}-\left[\frac{\left(\mathbf{33}-{\Psi }_{e,i}\right)\left({\theta }_i-{\theta }_{FC,i}\right)}{\mathbf{10}\left({\theta }_{s,i}-{\theta }_{FC,i}\right)}\right]& {\theta }_i \ge {\theta }_{fc,i}\\ \frac{A_i}{\mathbf{10}}{\theta }_i^{-B_i}& {\theta }_i<{\theta }_{fc,i}\end{array}$$

(17)

$$A_i=exp\left(ln\mathbf{33}+B_{i}ln{\theta }_{FC,i}\right)$$

(18)

$${\Psi }_{e,i}={\Psi }_{et,i}+\left(\mathbf{0.02}{\Psi }_{et,i}^{\mathbf{2}}-\mathbf{0.113}{\Psi }_{et,i}-\mathbf{0.70}\right)$$

(19)

$$\begin{array}{c}{\Psi }_{et,i}=\mathbf{21.674}\%S-\mathbf{27.932}\%Cl-\mathbf{81.975}\left({\theta }_{s,i}-{\theta }_{FC,i}\right)\\ +\mathbf{71.121}\%S\left({\theta }_{s,i}-{\theta }_{FC,i}\right)+\mathbf{8.294}\%Cl\left({\theta }_{s,i}-{\theta }_{FC,i}\right)\\ +\mathbf{14.05}\%S\%Cl++\mathbf{27.161}\end{array}$$

(20)

where for soil layer i, ${\mathrm{H}}_{\mathrm{m}}$ is the soil matric suction (m), ${\mathsf{\Psi }}_{\mathrm{e}}$ is the air entry suction (kPa), θ is the volumetric moisture content at layer i (m³ m⁻³), θ_s is the saturated volumetric water content (m³ m⁻³), θ_fc is the volumetric water content at field capacity (m³ m⁻³), %Cl is the clay content of the soil (%), and %S is the sand content of the soil (%).

Irrigation Subcomponent

The irrigation subcomponent determines when to irrigate the crop and how much water needs to be applied. The main program will call this subcomponent whenever the moisture depleted from the effective root zone goes below the specified depletion threshold during the entire cropping period. Every time the allowable percentage of available water is depleted, the irrigation subcomponent will notify the user when to apply the irrigation and estimate the amount needed to be supplied to bring the soil back to its field capacity. Unless effective rainfall occurs before the scheduled irrigation event, the farmer should irrigate the field as indicated in the plan suggested by the model.

The irrigation subcomponent also allowed simulation of non-uniform irrigation application where the estimation of the amount of water to be supplied is based on varying depletion thresholds and effective rooting zone set for each growth stage. Using the net irrigation estimated from the water balance in the main program, the irrigation subcomponent will calculate the gross irrigation volume needed by accounting for the efficiency of the mode of irrigation used, the total wetted area, and the surface area of the field. The duration at which the calculated volume is to be provided to the crop will be estimated using the application rate of the irrigation system employed.

Database Development and Management System

The program also offers a simple database equipped with climatic records from NAS-UPLB as well as with acceptable values of different soil and crop properties from published literature. The model also has a database management system that allows retrieval of available data and automatic storage of all new data inputted by the user.

Graphical User Interface

To provide easier and better interaction between the program and the farmer, a simple graphical user interface (GUI) was designed. The interface has five main features and a button as shown in Figure 3.

The first two features of the interface are the Field Information feature and the Weather Information feature. These are primarily intended for a user to provide key data about the field, crop, and weather. In the Field Information feature, users should provide information about the field such as field ID, crop and variety planted, the date of the emergence of the crop and its total growth duration, and the soil type. Weather-related data, on the other hand, may be provided by the user in the Weather Information feature of the program. Any rainfall occurrence during the growth duration of the crop can be inputted through this feature.

The next three features of the interface, on the other hand, were designed so that a user may have an easy and good grasp of the various useful information that can be derived from the results of the simulation. The Graphical Simulation feature provides users with a simple graphical presentation of the simulated scenario where the predicted days on which irrigation water is to be supplied to the crop can be easily determined. In this graph, the total available moisture that can be extracted by the crop is plotted using a blue line. The critical moisture level which will trigger an irrigation event is plotted using a red line (critical moisture), and the daily moisture depth variation throughout the cropping season is plotted using a green line. This will allow users to easily estimate intervals between irrigation events.

The user can also inspect the day-to-day status of available moisture in the field using the Water Balance Information feature of the program. In this feature, users can enter any date within the growing period of the crop and the program will give various information about that day including the current moisture status and moisture deficit, the need for irrigation, the volume of water needed, and the irrigation period. The feature is also equipped with a moisture status bar which allows the user to visually inspect the moisture level on the date entered. In the status bar, the available moisture that can be extracted by the crop is bounded by a blue bar. The critical moisture level which will trigger an irrigation event is plotted using a red bar, while the moisture level on the date entered is plotted using a green bar. The status bar helps a user to visually examine if the moisture level on the day inspected nears or exceeds the allowable or critical moisture level. Some inputs which should also be provided in this feature include the available discharge at which irrigation will be provided and the efficiency of the system in providing irrigation. The feature also houses the simulate button which can be activated by the user whenever the simulation needs to be updated (e.g., in the event of rainfall).

Users can also monitor the growth of the crop using the Phenological Stage feature of the program. In this feature, the growth of the crop estimated as a percentage of its total growth duration can be monitored using a horizontal bar divided into four parts using distinct colors, where each color represents a specific growth stage. In the case of this program, the horizontal bar colored mint green covers the initial stage of the crop, the mustard portion represents the crop development stage, the yellow-green bar covers the mid-season stage, and the dark green bar represents the late-season stage of the crop.

2.3. Model Calibration, Sensitivity Analysis, and Validation

The field capacity and permanent wilting point moisture contents of the soil are considered two of the key parameters estimated in the simulation model developed in this study. If in situ estimates of the properties are not available, the model will estimate these two values based on more readily available soil data, and the soil type. Although the equations used in the model were validated for a wide range of soil textural classes and were proven to give reasonable and accurate estimates, calibration is still recommended whenever particle size distribution data and field data on soil moisture characteristics are available. In calibrating the model, sensitivity analysis was performed to investigate how the changes in the value of the sand and clay content will affect the estimates of FC and PWP. Based on the results of the sensitivity analysis, calibration was then performed by adjusting the value of the sand and clay content within the acceptable limits to fit the soil moisture characteristic data obtained from the IRRI’s pF laboratory. The sand and clay content combination that provided the minimum discrepancy between the estimated and observed soil moisture characteristics were used in the simulation. Observed moisture level fluctuations from the pot experiments were used to calibrate the prediction of the model. For the validation of the model, four drip-irrigated corn plots (6 m × 10 m) were established at the experimental site, at CES Area B7. Scheduling of irrigation in the field was based on a non-uniform water application depending on the depth of the root zone and optimum MAD/critical moisture level for corn at different stages of growth. Soil moisture level in the field was monitored using moisture meters. The depths at which measurements were taken depended on the growth stage of the crop. Simulation performance was evaluated using statistical parameters including mean absolute error (MAE), root mean squared error (RMSE), normalized root mean squared error (NRMSE), and Nash–Sutcliffe efficiency (NSE).

Mean absolute error (MAE) calculates the mean of the residuals in a dataset or the absolute difference between observed and simulated data [23]. MAE is computed using Equation (21),

$$MAE=\frac{\mathbf{1}}{n}{\displaystyle \sum }_{i=\mathbf{1}}^n\left|y_i-x_i\right|$$

(21)

where n is the total number of data points, ${\mathrm{y}}_{\mathrm{i}}$ is the simulated value, and ${\mathrm{x}}_{\mathrm{i}}$ is the observed value.

Root mean squared error (RMSE) is a very common error index statistic used in numerical predictions. RMSE equal to zero indicates that the simulated data perfectly fits the observed data [23]. This statistic can be computed using Equation (22),

$$MSE=\sqrt{\frac{\mathbf{1}}{n}{\displaystyle \sum }_{i=\mathbf{1}}^n{\left(y_i-x_i\right)}^{\mathbf{2}}}$$

(22)

where n is the total number of data points, ${\mathrm{y}}_{\mathrm{i}}$ is the simulated value, and ${\mathrm{x}}_{\mathrm{i}}$ is the observed value.

Normalized root mean squared error (RMSE) is one of the extended metrics commonly used in numerical predictions. NRMSE equal to zero indicates that the simulated data perfectly fits the observed data [24]. This statistic can be computed using Equation (23),

$$NRMSE=\sqrt{\frac{\mathbf{1}}{n\left(x_{max}-x_{min}\right)}{\displaystyle \sum }_{i=\mathbf{1}}^n{\left(y_i-x_i\right)}^{\mathbf{2}}}$$

(23)

where n is the total number of data points, ${\mathrm{y}}_{\mathrm{i}}$ is the simulated value, and ${\mathrm{x}}_{\mathrm{i}}$ is the observed value.

Nash–Sutcliffe efficiency (NSE) is a “normalized statistic that determines the relative magnitude of the residual variance (“noise”) compared to the measured data variance (“information”)”. NSE describes how well the plot of observed and simulated data fits the line of equality [23]. NSE can be computed using Equation (24),

$$NSE=\mathbf{1}-\left[\frac{\frac{\mathbf{1}}{n}{{\displaystyle \sum }}_{i=\mathbf{1}}^n{\left(x_i-y_i\right)}^{\mathbf{2}}}{\frac{\mathbf{1}}{n}{{\displaystyle \sum }}_{i=\mathbf{1}}^n{\left(x_i-x_{mean}\right)}^{\mathbf{2}}}\right]$$

(24)

where n is the total number of data points, ${\mathrm{y}}_{\mathrm{i}}$ is the simulated value, ${\mathrm{x}}_{\mathrm{i}}$ is the observed value, and ${\mathrm{x}}_{\mathrm{mean}}$ is the mean of the observed data.

2.4. Model Simulations

Model simulations under varying rainfall scenarios were performed. In this study, two cases were examined: (1) decreased rainfall scenarios, and (2) increased rainfall scenarios. Observed rainfall depths in the field from 11 April 2021 to 24 June 2021 were obtained from the NAS-UPLB. The magnitude of the observed rainfall during the said cropping period was varied by ±25%, ±50%, ±75%, and ±100% to formulate the rainfall scenarios. Simulations for these scenarios under decreased and increased rainfall were performed and the capability and usefulness of the program in efficiently managing irrigation water under the different rainfall scenarios were investigated.

3. Results and Discussion

3.1. Model Calibration and Sensitivity Analysis

Using the particle size distribution data, sensitivity analysis was performed to investigate how adjusting the sand and clay content during the calibration of the model will affect the predicted value of FC and PWP. Three scenarios were investigated in the analysis. The first scenario involved changing the sand content within the acceptable limit set by the range of sand content of clay loam soil while using a constant value of clay content. The second scenario involved changing the clay content within the acceptable limit set by the range of clay content of clay loam soil while using a constant value of sand content. The last scenario involved changing both the sand and clay content of the soil within their acceptable limits.

The results of the sensitivity analysis revealed that adjusting the sand and clay contents of the soil did not result in an enormous change in the FC and PWP estimates. For example, when sand content was decreased by 40%, the FC estimate obtained from using the equation increased by around 10% only, while the PWP estimate decreased by around 1% only. Consequently, the increase in FC and the decrease in PWP resulted in about a 25% increase in the available water. The percentage change in output which is about 25% was relatively lower than the percentage change in input, which was 40%. It was also observed that when clay content was adjusted to a maximum percentage change of 20%, the percentage change in FC was lower than 10%, while it was nearly 20% in PWP. These changes caused the available water depth to increase to about 5% only. Similar to the observation when sand content was adjusted, the percentage change in output, which is about 5%, was comparatively lower than the percentage change in input, which was 20%. Furthermore, when both the sand and clay contents were adjusted, no enormous change in FC and PWP was observed, as shown in Figure 4. The results suggest that adjusting the values of sand and clay contents within the acceptable limit during calibration will not result in huge changes in FC and PWP estimates.

Observed moisture level fluctuations from pot experiments were used to calibrate the prediction of the model. The initial moisture content and the soil properties used in the calibration are shown in

Table 3. Soil properties used during model calibration.

Layer	Soil Depth (m)	Texture	Sand (%)	Clay (%)
1	0.3	Clay Loam	42.39	31.00

and

Table 4. Initial moisture content of the soil in the pot experiments.

Pot	Moisture Content, Volume Basis (m3/m3)
1	0.271
2	0.268
3	0.274

. Input parameters used in estimating the soil moisture characteristic were varied within the acceptable limits to fit the soil moisture characteristic data obtained from the laboratory and consequently the simulated depth generated by the program. The statistical parameters used to investigate the performance of the model in generating acceptable simulations include the coefficient of determination (R2), absolute error (AE), root mean square error (RMSE), and Nash–Sutcliffe efficiency (NSE). Figure 5 shows the plot of the simulated moisture depth and the observed moisture depth from the pot experiment.

Calibration resulted in a satisfactory matching between the observed and simulated depth as indicated by MAE = 8.00 mm and RMSE = 8.69 mm. In terms of NSE, some studies, as presented in a Summary Table by Moriasi et al. [23], would consider the results unsatisfactory (i.e., Ramanarayanan et al. [25], NSE > 0.4; Santhi et al. [26], NSE > 0.5; Singh et al. [27], NSE > 0.65), while they would be satisfactory according to other literature (Motovilov et al. [28], 0.36 ≤ NSE ≤ 0.75). In the case of this program, NSE = 0.3671 can still be described as satisfactory considering the significant effects of the geometric configuration of the container and the boundary condition of potted experiments in soil water flow. As shown in Figure 5, the simulation triggered irrigation either 2 days earlier or 2 days later. The advance or the delay in prediction can be associated with the effect of the container boundary, restricted root growth, and soil physical characteristics which may have been altered during potting. Although the effects of these factors were not verified in this study, several research studies have reported their influence on water flow. Moreover, the calibration data points used were also limited due to the maximum rooting depth that can be covered by the pots. Considering these contributing factors, the simulation can still be considered acceptable, and the NSE value is expected to be higher when the model is tested under field conditions.

3.2. Model Testing and Validation

Moisture level fluctuations observed from four (4) corn plots were used to validate the simulation of the program. The statistical parameters used to investigate the performance of the model in generating acceptable simulations include coefficient of determination (R2), mean absolute error (MAE), root mean square error (RMSE), and Nash–Sutcliffe efficiency (NSE).

Table 5. Initial moisture content of different soil layers used as input in validating the model.

Plot	Layer	Moisture Content, Volume Basis (m3/m3)
1	1	0.254
	2	0.263
	3	0.325
	4	0.330
2	1	0.256
	2	0.260
	3	0.325
	4	0.330
3	1	0.248
	2	0.256
	3	0.325
	4	0.330
4	1	0.254
	2	0.263
	3	0.325
	4	0.330

shows the initial moisture content of the different soil layers used as input in the simulations during validation while Figure 6 shows the plot of the simulated moisture depth and the observed moisture depth in the field.

The results show that the model exhibited better performance when tested under field conditions, as expected. Improved model performance is indicated by NRMSE values lower than 0.2 and higher NSE values ranging from 0.637 to 0.788. Moreover, the difference between MAE and RMSE remained minimal (<2 mm) which suggests that it is unlikely that huge errors in the simulation would be observed. The summary of the statistics obtained during model validation is shown in

Table 6. Summary of the statistics obtained during model validation.

Plot	Mean Absolute Error (mm)	Root Mean Square Error (mm)	Normalized Root Mean Square Error	Nash–Sutcliffe Efficiency
1	12.01	13.12	0.196	0.637
2	9.81	11.32	0.171	0.698
3	9.85	11.29	0.157	0.788
4	9.71	11.51	0.167	0.747

. The statistics reveal the capacity of the program to model soil moisture depth fluctuations observed in the field in an acceptable manner. Although the simulated and observed depths revealed a close variation, discrepancies between the two were noticed. For example, the difference between the predicted and the observed moisture depth is very close to or even lower than the RMSE for Plots 1, 3, and 4 at 23 days after emergence. A different case was observed in Plot 2 wherein the difference between the predicted and the observed moisture depth is relatively higher than the RMSE. Such a very high discrepancy may be attributed to the accuracy of the meter used to measure moisture content in the field. This was also observed in a number of studies that used probes for measuring moisture content [14]. Aside from the limitations of the device, the discrepancy may also be due to the spatial variability of moisture in the field. Since soil moisture content measurements in this study were point data and only a single point was measured per plot, it was very likely that highly variable moisture content measurements can be obtained between plots [29,30], particularly at the shallower layer of the soil. This was also observed by Li et al. [31] in potato fields, wherein the variability of moisture content tends to be higher at shallower layers of the soil, specifically when the moisture content in the field is relatively low.

Another observation made regarding the simulated and observed soil moisture depth is that even when a significant amount of rainfall to bring the soil back to field capacity occurred, the program consistently gave lower estimates of total available water. This is due to a very small underestimation of FC and small overestimation of PWP generated by the model using percentage sand and percentage clay as input variables. Although the statistics reveal good model performance, the simulation can still be improved if accurate estimates of FC and PWP are available. In this study, FC and PWP estimates were obtained from the pF laboratory. And since the program allows using laboratory-determined values as input, simulations using actual values were performed to further investigate the capability of the program. Figure 7 shows the plot of the simulated moisture depth using actual values of FC and PWP and the observed moisture depth in the field.

The results show that the model exhibited better performance when actual values of FC and PWP were used as input values. The summary of the statistics obtained from validating the model using actual values of FC and PWP is shown in

Table 7. Statistics obtained from validating the model using actual values of FC and PWP.

Plot	Mean Absolute Error (mm)	Root Mean Square Error (mm)	Normalized Root Mean Square Error	Nash–Sutcliffe Efficiency
1	5.76	6.83	0.102	0.901
2	7.14	8.56	0.129	0.827
3	6.87	8.12	0.113	0.891
4	6.46	8.11	0.118	0.874

. Improved model performance is indicated by lower NRMSE values ranging from 0.102 to 0.129, better NSE values ranging from 0.827 to 0.901, lower MAE values ranging from 5.76 mm to 7.13 mm, and lower RMSE values ranging from 6.83 mm to 8.56 mm.

3.3. Model Simulations under Various Rainfall Scenarios

To further demonstrate the capability and usefulness of the program as a tool for more efficient irrigation water management in corn production, model simulations under varying rainfall scenarios were performed. In this section, two cases are examined: (1) decreased rainfall scenarios, and (2) increased rainfall scenarios. The observed rainfall during the growth duration of the crop was varied by ±25%, ±50%, ±75%, and ±100% to simulate the soil moisture depth in the field under decreased rainfall scenarios. The initial moisture content in the field was set to trigger irrigation on the day of the emergence of the crop while all the other data needed to proceed with the simulation were similar to the validated model. The simulated moisture depths for each scenario under decreased and increased rainfall scenarios are shown in Figure 8 and Figure 9, respectively.

As shown in Figure 8, two possible cases may occur when rainfall is decreased. Either the interval between irrigation applications is shortened, or the number of irrigation applications is increased. For example, when the observed rainfall is decreased by 25%, the second irrigation event happened 4 days earlier or the interval between the first and second irrigation application was decreased from 18 days to 14 days. Further decrease in the observed rainfall not only shortened the interval between the first and second irrigation application but also increased the frequency at which irrigation water should be supplied to the crop. For instance, the number of irrigation applications was increased by one when rainfall was reduced by 50%. Further decrease would require the field to be supplemented with irrigation more frequently. Two more irrigation applications are needed when there is a 75% reduction in the observed rainfall, while three more irrigation applications should be provided when no rainfall occurred during the entire growth duration of the crop. The results highlighted the capability and usefulness of the program in helping farmers anticipate the amount of water required to supplement the crop under various decreasing rainfall scenarios. Moreover, the program will also allow users to maximize rainfall occurrences as the source of water to their field which can ultimately result in a significant amount of water savings. The irrigation schedule generated by the program can also be used to schedule fertilization or even fertigation, if applicable.

Two possible cases may also occur when there is increased rainfall (Figure 9). Either the interval between irrigation applications is lengthened, or the number of irrigation applications is reduced. For example, when rainfall is increased by 25%, the second irrigation application will happen 3 days later or the interval between the first and second irrigation application is increased from 18 days to 21 days. Further increase in the observed rainfall not only lengthened the interval between the first and second irrigation application but also reduced the frequency at which irrigation water should be supplied to the crop. For instance, the number of irrigation applications decreased by one when rainfall was increased by 50%. Further increase, however, would no longer decrease irrigation frequency. This highlights the importance of accounting for the distribution of rainfall throughout the crop production process. The frequency of rainfall occurrences is as important as the magnitude of rainfall received by the crops. Moreover, the result also highlights one of the limitations of the program as it does not provide information on excess moisture considering that the simulated moisture depth is only the water readily available to the crops. Hence, the program is not useful for drainage purposes. The results, however, still suggest that a significant amount of water savings can be achieved when rainfall is properly accounted for in managing water in the field.

4. Conclusions

The study was able to address the need for a new water management technique that will reduce water use in corn production systems while maintaining optimum yields through the development of a spreadsheet-based water balance model. The program developed will capacitate corn farmers in managing water in the field as it functions as a practical decision tool in determining the correct amount of irrigation water to be applied during critical growth stages of the crop. Although further improvements can still be made to the developed program, statistics revealed that the model exhibited good performance as it simulated soil moisture depth throughout the growing period of the crop in an acceptable manner. The capability and usefulness of the program as a tool for more efficient irrigation water management in corn production were also demonstrated by model simulations under varying rainfall scenarios, the results of which proved to be useful in maximizing the benefit from rainfall occurrences as the source of water for their corn crop. The simulations under increased rainfall scenarios emphasized that a significant amount of water savings can be achieved when rainfall is properly accounted for in managing water in the field as it can decrease the number of irrigation applications or lengthen the interval between applications. It was also revealed that the frequency of rainfall occurrences is as important as the magnitude of rainfall received by the crops. Model simulations under decreased rainfall scenarios proved that the model is useful, particularly during seasons of longer dry periods. The results highlight the capability and usefulness of the program in helping farmers anticipate the amount of water to supplement the crop under various decreasing rainfall scenarios as it provides information on how decreasing rainfall will increase the number of irrigation applications needed or shorten the interval between irrigation applications. In general, the program offers a simple interface, requires minimum data input yet provides satisfactory and useful results for the irrigation water management of corn. The wide-ranging user-friendliness and simplicity of the model developed in this study can pave the way to eliminating the barriers which cause farmers to resist advancements in their farming practices as the model can easily be used not only by researchers and scientists but also by farmers, especially those with basic knowledge of spreadsheets.