Optimizing furrow irrigation performance with WinSRFR software through field experiments and simulations in sunflower cultivation

Abstract

Surface irrigation continues to dominate agricultural practices in India; however, its effectiveness is often limited by deep percolation losses and non-uniform water distribution. This study, conducted during 2023–2024 and 2024–2025 at ICAR-CIAE, Bhopal, India aimed to evaluate and optimize furrow irrigation methods for sunflower cultivation using WinSRFR simulation software. Continuous Furrow Irrigation (CFI) and Alternate Furrow Irrigation (AFI) were tested under varying furrow lengths (45, 55, 65 m) and inflow rates (1.0, 1.5, 2.0 lps) across three crop growth stages. Field data on infiltration, advance and recession times, and soil moisture were collected and used to calibrate the model. Simulation results revealed that AFI with a 65 m furrow length, 1.0 lps flow rate, and 80-min cutoff time achieved the highest performance: 87% Application Efficiency (AE), 91% Distribution Uniformity (DU_lq), and only 15% Deep Percolation (DP). In contrast, CFI under optimal conditions (1.1 l ps, 64.4 min) reached 71% AE. Model validation showed high agreement with observed values (R² > 0.93, RMSE < 1.4%, p < 0.001). Sensitivity analysis identified inflow rate and cutoff time as the most influential variables. Cross-season validation confirmed the robustness of optimized parameters under real field conditions. These findings offer practical, scalable recommendations for improving water productivity in Indian furrow irrigation systems using simulation-assisted planning.

Introduction

Water scarcity remains one of the most pressing challenges in global agriculture, especially in arid and semi-arid regions where water availability directly determines crop productivity. Globally, irrigated agriculture accounts for only 20% of cultivated land yet contributes approximately 40–45% of total food production¹. In Asia, over 90% of irrigated farms still rely on traditional surface irrigation methods², highlighting the urgent need for improving these systems. Surface irrigation, while simple and low-cost, exerts a substantial impact on catchment-scale water resources and often leads to inefficiencies due to deep percolation losses and uneven water distribution^3,4. Improving the design and management of surface irrigation systems has been shown to enhance irrigation efficiency, increase water productivity, and reduce agrochemical contamination risks^5,6,7.

In India, surface irrigation remains dominant, accounting for over 85% of the total irrigated area despite its relatively low water use efficiency of 30–40%⁸. Furrow irrigation, a commonly practiced surface irrigation method, channels water between crop rows, providing improved aeration and soil moisture control for row crops⁹. However, challenges such as deep percolation losses, uneven infiltration, and poor water distribution efficiency persist, particularly under clayey soil conditions^10,11. Factors such as stream size, furrow length, slope, and cut-off time critically influence irrigation performance^7,12,13. Previous studies have indicated that surface irrigation can achieve efficiency levels similar to those of well-managed sprinkler systems, as long as these systems are operated and managed properly¹⁴. This highlights the potential for similar advancements in India through the adoption of modern surface irrigation techniques and technologies. In India, fragmented landholdings further complicate irrigation management, often resulting in small plots irrigated individually and increasing water losses⁸.

Continuous furrow irrigation (CFI) and alternate furrow irrigation (AFI) are among the most commonly adopted furrow irrigation methods in many developing countries due to their simplicity, cost-effectiveness, and minimal infrastructure requirements^15,16. However, CFI often results in significant deep percolation losses, particularly in clayey soils where infiltration is slow and uneven^10,12. In contrast, AFI, where water is applied to every alternate furrow has emerged as a promising strategy for improving water productivity and saving water without adversely affecting crop yield^17,18. Despite these advantages, limited research has directly compared the hydraulic performance of CFI and AFI under variable field conditions and across multiple irrigation events using simulation-based approaches.

Simulation in surface irrigation systems involves mathematically modeling the hydraulic behaviour of water as it moves across a field¹⁹. While parameters such as field slope and length are challenging or costly to modify once the system is in place, variables such as inflow timing, cut-off time, inflow rate, and desired application depth can be optimized through the use of simulation models. This optimization is accomplished through computer models that utilize mathematical equations, particularly the Saint–Venant equation. Various surface irrigation simulation models have been developed, with SRFR²⁰, SIRMOD²¹, and WinSRFR²² being the most widely used globally. Advances in simulation models, particularly the WinSRFR software developed by the USDA Agricultural Research Service²², have enabled more precise optimization of surface irrigation parameters. WinSRFR uses simplified hydraulic equations to simulate water flow, predict infiltration, and optimize inflow rates and cut-off times²³. Studies have validated its high accuracy in simulating advance-recession curves and infiltration dynamics under varying field conditions^3,4. However, most prior research has focused either on experimental assessments or standalone simulations under fixed conditions, with limited integration of both approaches for comprehensive optimization^13,24,25. Furthermore, few studies have validated these models using field data across multiple irrigation events, especially in sunflower cultivation under Indian semi-arid clayey soils. The integration of WinSRFR simulation with empirical field data for such comparative analysis has not been adequately explored.

The novelty of this research lies in its dual methodological framework, merging empirical field assessments with advanced simulation modeling for both CFI and AFI. Unlike previous studies that rely on theoretical models or single-season trials, this work delivers a validated, multi-event analysis applicable to real farming scenarios. The results will directly aid farmers, agronomists, and water managers in selecting optimal irrigation strategies for improved water productivity and sustainability in regions with similar agro-ecological conditions.

This study addresses these gaps by systematically combining field experiments with WinSRFR modeling to optimize furrow irrigation performance in sunflower cultivation. Conducted in 2023–2024 and 2024–2025 in PFDC, ICAR-CIAE, Bhopal. The research evaluates CFI and AFI under different furrow lengths and flow rates, measuring soil infiltration characteristics and hydraulic performance indicators. The primary objectives are to (i) calibrate the WinSRFR model using field data, (ii) simulate and optimize inflow rates, cutoff times, and furrow dimensions, and (iii) perform external field validation in the following irrigation season to confirm model robustness. This integrated approach provides practical, field-validated recommendations for improving Application Efficiency (AE), Distribution Uniformity (DU_lq), and minimizing Deep Percolation (DP).

Materials and methods

Research area

An experiment was conducted at the Precision Farming Development Centre (PFDC) of the ICAR-Central Institute of Agricultural Engineering, Bhopal, Madhya Pradesh, in central India, to evaluate and enhance the performance of CFI and AFI methods. The experimental site lies between latitudes 23° 18′ 22′′ and 23° 20′ 00′′ N, and longitudes 77° 24′ 45′′ and 77° 25′ 24′′ E, with an average elevation of 490 m above mean sea level. The region has a humid subtropical climate, characterized by cool, dry winters, hot summers, and a humid monsoon season, with an average annual rainfall of 1,146 mm. Summer typically starts in late March and continues until mid-June, with temperatures frequently exceeding 40 °C (104°F) in May. Figure 1 shows the geographical location of the study area.

Fig. 1

Location map of the study area. The map was created and processed using QGIS 3.12 software (available at https://qgis.org/download/).

Soil type

Soil texture was analyzed using the hydrometer method, and the textural class was categorized according to the USDA system. Soil bulk density was measured using the core cutter method. Soil moisture at field capacity (FC) and permanent wilting point (PWP) were determined at pressure heads of 33 kPa and 1500 kPa, respectively, using a pressure plate apparatus and pressure membrane. Soil EC and pH values were measured using a digital EC-pH meter. Lastly, the hydraulic properties of the soil were determined using standard field methods. The results of the soil physicochemical and hydraulic properties are summarized in (Table 1).

Table 1 Soil Physical and hydraulic properties of the experimental farm.

Field experiment details

During the 2023–2024 and 2024-25 sunflower growing season, a field experiment was conducted with two furrow irrigation methods: CFI, in which each furrow received water, and AFI, where water was supplied to every other furrow alternatively. The study tested different combinations of furrow length and discharge rates (Fig. 2). Due to the variations in soil infiltration characteristics throughout the cropping period, three separate evaluations were conducted.: at the beginning (1st irrigation) when the soil was nearly bare shortly after sowing, in the middle (3rd irrigation) during the full vegetative stage, and at the end (5th irrigation) during the crop’s maturity stage. Parameters such as inflow rate, cutoff time, advance time, recession time, and infiltration depth were recorded for each irrigation event. The experimental site was prepared through deep ploughing with an MB plough, followed by a cultivator and rotavator for fine ploughing. A laser-guided land leveller was used to create a uniform longitudinal and cross-sectional slope. Furrows were then shaped using a tractor-operated ridge and furrow former.

Fig. 2

Layout of the field experiment.

Data collection

In each treatment, the data were collected from the middle furrow by leaving two adjacent furrows as buffer furrows to eliminate potential lateral flow measurement errors. Stakes were placed along the furrows at fixed intervals of 10 m, with the final observation point marking the actual remaining furrow length, as long as the remaining length was within 10 m. Advance times were noted when the waterfront reached each stake, while recession times were recorded once the water had completely infiltrated or disappeared at the observation points. The experiments were carried out using three different furrow lengths: 45 m, 55 m, and 65 m, with a furrow width of 0.60 m and a longitudinal slope of 0.002 m/m. All experiments were performed under closed-end boundary conditions. Uniform and constant flow rates of 1.0, 1.5, and 2.0 lps were supplied through the PVC pipe network. The discharge rate of each furrow stream was controlled using a 32 mm valve installed in the PVC pipeline at intervals of 1.2 m. A magnetic brass water meter (Chambal Magnetics, N.B. Industries (Meter) Pvt Ltd, India) was fitted to the pipeline at the inlet to measure the total volume of water used. To ensure uniformity in data collection, the same layout was used for each treatment. The water supply for the experiments was drawn from a sump with a capacity of 1,00,000 L. The required constant discharge was maintained by adjusting the pump pressure and opening area of the supply valve. Irrigations were conducted in sequence, as the system could not handle all experiments simultaneously. The maximum non-erosive flow rate for the furrows was estimated using an empirical formula (Eq. 1) provided by Michael²⁶.

$${\text{q}}_{\text{m}}=\frac{0.60}{\text{s}}$$

(1)

The maximum non-erosive streamflow rate (q_m), in liters per second (lps), was determined using the empirical relationship, where q_m is influenced by the longitudinal slope of the furrow (s), expressed as a percentage. Based on this equation, the maximum non-erosive inflow rate of 3 lps for the experimental furrows was determined.

The experimental field was prepared for sunflower cultivation (KBSH 44) using a furrow irrigation system. Sunflower seeds were sown in the fourth week of December 2023, with harvesting in mid-May 2024, similar cultivation practices were adopted in 2024-25. The geometric dimensions of the experimental furrows were measured using a profilometer fabricated in the research workshop (Fig. 3). These dimensions were recorded at several points within the same furrow during each evaluation, and average values were used. Furrow cross-sectional data were subsequently employed to compute infiltration and shape parameters via volume balance analysis. Irrigation continued for approximately 10 min after the water reached the furrow ends.

Fig. 3

Cross-section of the furrow.

Inflow rates for the furrows were measured using digital 2-in cut-throat flumes (Fig. 4). An ultrasonic sensor was attached to each flume to capture inflow variations. Calibration of the flumes was performed using volumetric measurements. The target irrigation depth was established at 75 mm to ensure uniform seed germination and crop establishment, in line with current regional practices. Initial soil water content (SWC) was measured before the start of the experiments and 48 h after each irrigation event. Soil samples were collected from a soil profile up to 100 cm depth at five intervals (0–20 cm, 20–40 cm, 40–60 cm, 60–80 cm, and 80–100 cm) at observation points along the furrow lengths for each treatment. The volumetric SWC was derived to estimate the depth of infiltrated water by multiplying with the corresponding soil depth (20 cm) and bulk density. Finally, performance indicators were computed based on the stored water depth in the soil profile.

Fig. 4

The basic equipment used during the experiment and field measurement.

WinSRFR software

This study utilized WinSRFR 5.1, the most recent version of the WinSRFR series²⁷. WinSRFR software utilizes the numerical solution of the Saint–Venant equation (Eqs. 2–3) and integrates earlier tools such as SRFR²⁰, BASIN²⁸, and BORDER²⁹. The software allows for calculations using zero-inertia and kinematic wave models through numerical solution methods²⁰. WinSRFR is structured into four main components: event analysis, performance analysis, physical design, and simulation. The input data for winSRFR software is provided in Table 2. Initially, the event analysis module was utilized to calibrate the software and assess the actual irrigation performance in the sunflower field. Following this, the simulation module was employed to model the advance and recession trajectories, as well as the irrigation performance, for each treatment during each irrigation event. The outputs of these simulations included advance and recession trajectories, final infiltration profiles, depth hydrographs, and performance summary. Based on the recommendations of Bautista et al.³⁰, the zero-inertia model was selected due to the closed-end boundary conditions and the low slope of the furrows²⁷.

Table 2 Input Parameters for numerical simulation of furrow.

This model is grounded in the governing principles of the Saint–Venant equation;

$$\frac{\partial y}{\partial t}+\frac{\partial q}{\partial x}+\frac{\partial z}{\partial t}=0$$

(2)

$$\frac{\partial q}{\partial x}+\frac{2q\partial q}{y\partial x}+gy\frac{\partial y}{\partial x}=gy\left({S}_{o}-{S}_{f}\right)$$

(3)

In this equation, y represents the depth of flow in meters, q represents the discharge per unit width in m³/s, z is the infiltrated volume per unit length (m³/m), g represents the acceleration due to gravity in m²/s, ${S}_{o}$ (m/m) denotes the field slope and ${S}_{f}$ (m/m) denotes the energy gradient.

The zero-inertia model used in this research work, where the above equation becomes

$$\frac{\partial y}{\partial x}=\left({S}_{o}-{S}_{f}\right)$$

(4)

The selection of input parameters in WinSRFR was based on a combination of field measurements, literature benchmarks, and iterative calibration. Furrow geometry (length, width, slope) was obtained using a pin profilometer and GPS survey, while the average required irrigation depth (75 mm) was derived from regional agronomic recommendations for sunflower during the early growth stage. Manning’s roughness coefficient (η) was estimated through field observations of flow depth and wetted perimeter and cross-verified using established values reported for similar soil and crop conditions^16,31. The cut-off time and inflow rate combinations used for simulation were selected based on the maximum allowable non-erosive flow and regional best practices. These criteria ensured that the model inputs realistically represented field conditions and enabled accurate prediction of hydraulic performance indicators. Research methodology flowchart (Fig. 5) illustrating the sequential steps of the study, including field setup, experimental design, data collection, model calibration, simulation, validation, cross-validation, optimization, and final recommendations. The flowchart highlights the integration of field experiments with the WinSRFR model to evaluate and optimize furrow irrigation performance under Continuous Furrow Irrigation (CFI) and Alternate Furrow Irrigation (AFI) methods. Color coding is used to distinguish between different methodological phases for clarity.

Fig. 5

Flow chart of the methodology.

Estimation of infiltration parameters

In this study, the infiltration depth was estimated using the modified Kostiakov–Lewis equation³². The equation is generally represented as follows:

$${\text{Z}} = {\text{k t}}^{\alpha } + {\text{ f}}_{0} {\text{ t}}$$

where, Z represents the cumulative infiltration (mm), t is the infiltration opportunity time (hours), f₀ represents the basic infiltration rate (mm/hr), and k and α are empirical coefficients, with α is dimensionless and k is measured in mm·hr^-α.

The base infiltration rate f₀ is determined according to Walker and Skogerboe³³ as follows:

$${\text{f}}_{0}=({\text{Q}}_{\text{in}}-{\text{Q}}_{\text{out}})/\text{L}$$

where Q_in and Q_out are the inflow and outflow rate (m³ min⁻¹), and L is the length of the furrow (m), respectively.

The two-point method was used to determine the coefficients α and k³³. In the two-point method, the advance curve is determined using Eq. given by Eliot and Walker³⁴.

$$x= {pt}^{r}$$

where p and r are the fitting parameters, and t is the time from the start of inflow to reach station x. Finally, the other two parameters (α and k) are determined using volume balance Eqs.

$$\upalpha = \frac{{{\text{log}}\left( {{\text{V}}_{{\text{L}}} /{\text{V}}_{{0.5{\text{L}}}} } \right)}}{{{\text{logt}}_{{\text{L}}} /{\text{t}}_{{0.5{\text{L}}}} }}$$

$${\text{K}} = { }\frac{{{\text{V}}_{{\text{L}}} }}{{\upsigma _{{\text{Z}}} {\text{t}}_{{\text{L}}}^{\upalpha } }}$$

$${\text{V}}_{{0.5{\text{L}}}} = \frac{{{\text{Q}}_{0} {\text{t}}_{{0.5{\text{L}}}} }}{{0.5{\text{L}}}} -\upsigma _{{\text{y}}} {\text{A}}_{0} - \frac{{{\text{f}}_{0} {\text{t}}_{{0.5{\text{L}}}} }}{{1 + {\text{r}}}}$$

$${\text{V}}_{{\text{L}}} = \frac{{{\text{Q}}_{0} {\text{t}}_{{\text{L}}} }}{{\text{L}}} -\upsigma _{{\text{y}}} {\text{A}}_{0} - \frac{{{\text{f}}_{0} {\text{t}}_{{\text{L}}} }}{{1 + {\text{r}}}}$$

$$\sigma_{Z} = \frac{{\alpha + r\left( {1 - \alpha } \right) + 1}}{{\left( {1 + \alpha } \right)\left( {1 + r} \right)}}$$

where σ_y is the surface profile shape factor (0.77); σ_z is the subsurface profile shape factor; A₀ is the wetted area at the upstream (m²) and t_0.5L and t_L are the advance times (min) at two points, x₁ = 0.5L and x₂ = L, respectively.

Ismail and Depeweg³⁵ demonstrated its effectiveness in simulating block-ended furrow irrigation for short furrows (< 100 m). However, one limitation of this method is its reliance on two data points, which makes it highly sensitive to the measurement of advance times. To improve accuracy, Bautista et al.²² recommended collecting additional data points during field measurements. In this study, data were collected at multiple stations along the furrows and recorded until the end of the recession phase to improve the accuracy of the two-point method. Given the sensitivity of furrow irrigation performance to infiltration parameters, calibration was required. The Kostiakov–Lewis coefficients were adjusted using the Event Analysis module and the Merriam and Keller³⁶ method within the WinSRFR software. Infiltration parameters were outputs from the Event Analysis and could not be directly configured. Calibration was performed through trial and error, where the parameter α was entered as input and the infiltration parameters were adjusted within an acceptable range (RMSE < 10%). The agreement between the measured and simulated advance-recession times, along with the infiltrated water depth, was utilized to determine the optimal combinations of α, k, and f₀.

Manning’s roughness coefficient (η)

Manning’s equation³⁷ was applied to estimate the roughness coefficient, assuming uniform flow and that the flow depth reached the normal depth. The cross-sectional area and wetted perimeter of the experimental furrows were determined using a pin profilometer, following the method described by Walker and Skogerboe³³. Measurements were taken at three locations along the furrows: the beginning, middle, and end. Manning’s equation was then applied to estimate the mean value of η once the infiltration rate reached a steady-state condition. The value of η is considered a function of the channel surface characteristics and is assumed to be independent of flow rate and depth. In this study, the same η value was assumed for similar furrow discharges, regardless of furrow length or irrigation method. The calculation process followed the method outlined by Mailapalli et al.¹⁶, which was adopted for use in this study.

Performance evaluation of WinSRFR software

The accuracy of the WinSRFR software in predicting advance and recession times was evaluated by comparing the simulated results with actual field measurements. In this study, three statistical criteria viz., Root Mean Square Error (RMSE), The Willmott Agreement Index (d), and Nash–Sutcliffe Efficiency (NSE) were used to assess the accuracy of the software. The formulas are provided as follows:

Criteria	Formula	Range	References
RMSE	$RMSE= \sqrt{\frac{1}{n}\sum_{i=1}^{n}{{(X}_{obs,i}-{X}_{sim,i})}^{2}}$	0 to infinity	Willmott and Matsuura38
Willmott Agreement Index (d)	$d=1-\frac{{\sum_{i=1}^{n}({X}_{sim, i}-{X}_{obs,i})}^{2}}{\sum_{i=1}^{n}{(\left|{X}_{sim,i}-{\overline{X} }_{obs,i}\right|+\left|{X}_{obs,i}-{\overline{X} }_{obs}\right|)}^{2}}$	0 to 1	Willmott39
NSE	$NSE=1-\frac{{\sum_{i=1}^{n}({X}_{obs,i}-{X}_{sim,i})}^{2}}{{\sum_{i=1}^{n}({X}_{obs,i}-{\overline{X} }_{obs})}^{2}}$	Negative infinity to 1	Nash and Sutcliffe40

where X_sim,i is the simulated value at time step i, X_obs,i is the measured value at time step i, ${\overline{\text{X}} }_{\text{obs}}$ is the mean of measured data, and n is the number of measurements.

RMSE offers a clear measure of how well a model fits the data by quantifying the average magnitude of errors between the measured and simulated values. A lower RMSE indicates better model performance, as it suggests that the simulation outputs are close to the observed data. RMSE provides a measure of the average magnitude of the prediction errors, but it is sensitive to outliers and does not indicate the direction of the deviation. Willmott’s d-index is a widely used complementary index that accounts for both systematic and unsystematic errors; however, it tends to exaggerate model performance in some cases due to its reliance on squared differences. NSE evaluates how well the plot of observed versus simulated data fits the 1:1 line, but it can produce misleadingly low values when the observed mean is close to zero or when large errors occur. While these metrics collectively provide insight into model performance, they do not capture uncertainty or confidence in predictions.

To ensure the robustness of model validation, paired t-tests were applied to assess whether statistically significant differences existed between observed and simulated values for application efficiency (AE), deep percolation (DP), and distribution uniformity (DU_lq). The calculation of 95% confidence intervals provided additional insights into the precision of the mean differences, enhancing interpretability and reliability of the results. Previous research has demonstrated that integrating these statistical methods with simulation models like WinSRFR strengthens the accuracy of model validation, particularly under variable inflow rates and field conditions^{15,16,23,25,41}.

Performance indicators of furrow irrigation

The assessment of CFI and AFI methods was carried out using three primary performance indicators: Application Efficiency (AE), Deep Percolation (DP), and Low-Quarter Distribution Uniformity (DU_lq)^3,6,22. These indicators provide insights into both the effectiveness and uniformity of irrigation practices. AE refers to the proportion of applied water that is effectively utilized by the crop, with higher AE values signifying more efficient water use. DP indicates the percentage of water that percolates below the root zone and is not available for crop uptake, often indicating water loss. DU_lq​ evaluates how evenly water is distributed across the field, with a focus on the lowest-watered areas. Higher DU_lq values indicate more uniform water distribution. It is essential to recognize that the performance of irrigation methods is generally measured using the DU_lq, whereas the efficiency of irrigation management is typically evaluated through AE or the fraction of beneficial water usage³.

The performance indicators were calculated using the following formulas (Eqs. 6–8):

$$AE= \frac{{Z}_{req}L}{Q. {T}_{co}}100\%$$

(6)

$$DP= \frac{{V}_{z}-{Z}_{req}L}{Q.{T}_{co}}100\%$$

(7)

$${DU}_{lq}=\frac{{Z}_{lq}}{\overline{Z} }$$

(8)

where, Z_req = Depth of water required in the root zone, Z_lq = Average water depth infiltrated in the 25% least irrigated area, $\overline{Z }$ = Average water depth infiltrated in the entire furrow, Q = furrow discharge rate, L = Length of furrow, T_co = cutoff time.

Optimization of furrow irrigation performance

The WinSRFR software facilitates optimization using operational analysis module. The cutoff time and inflow rate were optimized while maintaining constant furrow length, width, and slope. The optimal T_co was determined as the shortest duration that yielded the highest irrigation performance. AE, DU_lq and DP were determined for each optimized cutoff time and inflow rate for selected furrow lengths, and these performance metrics were compared with the existing scenario.

Bautista et al.²², explained the calculation of potential application efficiency (AE_pot), which represents the maximum efficiency achievable when the average infiltrated and stored water depth equals the required depth. This value reflects the potential performance of an irrigation method under optimal management, minimizing deep percolation losses and ensuring proper flow rates and application times. In general, an optimal combination of discharge (q) and cutoff time (T_co) maximizes both application efficiency and distribution uniformity. One method for determining optimal conditions involves a minimum threshold for distribution uniformity can be set, and the corresponding combination of q and T_CO that maximizes application efficiency can then be chosen²⁷. In this study, a distribution uniformity of 90% or higher was classified as high, with a target of 99% for requirement efficiency (RE) to achieve optimal application efficiency. Iso-performance curves for application efficiency and distribution uniformity were examined, assuming that the minimum infiltration depth equalled the required depth (D_min = D_req).

Sensitivity analysis of furrow irrigation parameters

To ensure the robustness and reliability of the WinSRFR model outputs, a sensitivity analysis was conducted focusing on the key input parameters: inflow rate, cutoff time, and infiltration characteristics (Kostiakov-Lewis parameters k, α, and f₀), as well as Manning’s roughness coefficient (η). These parameters were varied systematically to assess their influence on key performance indicators: Application Efficiency (AE), Deep Percolation (DP), and Distribution Uniformity (DU_lq).

For this analysis, the input parameters were independently altered by ± 25% and ± 50% from their baseline calibrated values, while other variables were held constant, following the methodology suggested by Mehri et al.⁴². Sensitivity coefficients (SC) were calculated to quantify the relative impact of each parameter on the output variables. The sensitivity coefficient (SC) was derived using the following formula⁴³:

$$\text{SC}=\frac{\left(\Delta \text{O}/\text{O}\right)}{\left(\Delta \text{I}/\text{I}\right)}$$

where ΔO and ΔI are the changes in output and input parameters, respectively, and O and I are their respective average values. The influence level of each parameter was categorized as High (SC ≥ 0.7), Moderate (0.3 ≤ SC < 0.7), or Low (SC < 0.3), following threshold criteria used in previous modeling studies^42,44. The analysis aimed to identify the most critical parameters affecting model performance to support accurate calibration and reliable field application.

Cross-validation of optimized parameters

To evaluate the robustness and practical applicability of the optimized furrow irrigation parameters derived from the WinSRFR model, an external cross-validation exercise was performed during the subsequent irrigation season (2024–2025). Following the initial calibration and optimization, the optimal discharge rates, cutoff times, and furrow dimensions identified from simulations during 2023–2024 were implemented under actual field conditions. For the AFI method, a discharge rate of 1.0 lps, a cutoff time of 80 min, and a furrow length of 65 m were applied. For CFI, the optimized parameters included a discharge rate of 1.1 lps, a cutoff time of 64.4 min, and the same furrow length of 65 m.

Field data collection was carried out during the 1st, 3rd, and 5th irrigation events, with three replications for each method and furrow treatment. Observations included advance and recession times, infiltration opportunity time, and soil moisture content, emulating the parameters measured during the calibration phase. The AE, DP, and DU_lq were computed using both observed field data and WinSRFR simulated outputs based on the optimized settings. The accuracy of cross-validation was further quantified using RMSE, R², and Pearson correlation coefficients for each performance indicator. This cross-validation approach provided an independent assessment of the model’s predictive capabilities under actual field conditions, confirming the reliability of the optimized irrigation parameters for practical application.

Results and discussion

Variation of infiltration parameters, cutoff time, and manning’s roughness coefficient

Table 3 presents the values of infiltration parameters, cutoff time, and Manning’s roughness coefficients for the CFI and AFI methods. The values of the infiltration parameter α varied across irrigation events and flow rates. For the 1st irrigation event, the α values ranged from 0.17 to 0.49 for CFI and 0.29 to 0.49 for AFI. In the 3rd irrigation event, α values ranged from 0.30 to 0.49 for CFI and 0.33 to 0.55 for AFI, while in the 5th irrigation event, the range was from 0.23 to 0.55 for CFI and 0.34 to 0.39 for AFI. When analysing the impact of inflow rates (1 lps, 1.5 lps, and 2 lps), the mean values of α for CFI were 0.313, 0.393, and 0.44, respectively. For AFI, the corresponding mean values of α were 0.354, 0.375, and 0.381. This trend indicates that the α parameter increases with increasing flow rates. During the growing season, the mean values of α for CFI were 0.350, 0.422, and 0.372 in the 1st, 3rd, and 5th irrigation events, respectively. For AFI, the mean values were 0.364, 0.375, and 0.377 for the same irrigation events. This shows that α increased during the 3rd irrigation and decreased again in the 5th irrigation for both CFI and AFI. Xu et al.⁴⁵ noted that changes in soil structure, caused by cultivation, irrigation, and rainfall, periodically disturb and alter infiltration parameters, including α. Furman et al.⁴⁶ proposed that (α) does not necessarily need to be treated as a variable parameter, though multiple studies have indicated that factors like soil bulk density and soil water content can affect α^45,47.

Table 3 Results of infiltration parameters and Manning’s roughness coefficient under different treatment.

The temporal variation of the basic infiltration rate (f₀) during the irrigation events showed distinct differences between CFI and AFI. In the first irrigation event, f₀ ranged from 2.9 to 3.3 mm/h for CFI and 4.0 to 4.3 mm/h for AFI. By the third irrigation event, f₀ increased to 3.5 to 4.2 mm/h for CFI and 4.1 to 4.5 mm/h for AFI. However, during the fifth irrigation event, f₀ decreased to 2.7 to 3.0 mm/h for CFI and 3.1 to 3.8 mm/h for AFI. The infiltration coefficient (k) also varied over time. In the first irrigation event, k values ranged from 31.99 to 64.54 mm/hr^α for CFI and 56.18 to 118.63 mm/hr^α for AFI. In the third irrigation event, k values increased to 49.35 to 81.33 mm/hr ^α for CFI and 65.29 to 126.76 mm/hr^α for AFI. By the fifth irrigation event, k ranged from 36.20 to 68.23 mm/hr^α for CFI and 60.79 to 107.18 mm/hr^α for AFI. Mazarei et al.⁴⁸ also observed an increase in k values as the growing season advanced. This temporal variation in k can be linked to factors such as Manning’s roughness coefficient and canopy cover⁴⁵. Dixon et al.⁴⁹ noted that higher k values are typically associated with micro-rough and macro-porous soil surfaces, or conditions that enhance gravitational influence on infiltration. The higher k values observed in the first irrigation event indicate that the soil was able to absorb water at a faster rate, likely due to the dry initial condition of the soil pores¹⁰. Conversely, smaller k values are linked to smoother, micro-porous surfaces where capillary forces dominate the infiltration process.

(Table 3) presents the measured values of Manning’s roughness coefficients (η) under different irrigation treatments. The first irrigation event, conducted under nearly bare soil conditions a few days before germination, showed η ranging from 0.017 to 0.023 for CFI and 0.014 to 0.020 for AFI. During the 3rd and 5th irrigation events, η value were evaluated only for the 45 m furrow length across all three discharge rates for both CFI and AFI, and the same values were applied to the 55 m and 65 m furrows for corresponding discharges. As the growing season progressed, crop growth and canopy cover increased, creating additional drag forces for the inflow rate, which led to an increase in Manning’s roughness. Previous research has likewise shown that canopy cover and crop residues can increase Manning’s roughness coefficient during the irrigation season^16,31,50. In the 3rd irrigation event, the η values ranged from 0.083 to 0.117 for CFI and 0.083 to 0.115 for AFI, while in the 5th irrigation event, the η values ranged from 0.048 to 0.062 for CFI and 0.043 to 0.059 for AFI. Research by Zhang et al.⁵¹, Etedali et al.⁵², Nie et al.⁵³, Anwar et al.⁵⁴, and Mazarei et al.⁶ reported that values of η typically range from 0.014 to 0.121, aligning with the results of this study. They also noted that Manning’s roughness tends to rise during the growing season due to temporal variability. Supporting these observations, Mailapalli et al.¹⁶, and Xu et al.⁴⁵, Mazarei et al.^6,48, reported that the lowest value of Manning’s roughness occurs at the start of the irrigation season, gradually increases until mid-season, and then declines toward the end of the season, aligning with the findings of the current study. As presented in Table 3, the highest Manning’s roughness values during all irrigation events were recorded at an inflow rate of 1 lps, with η decreasing as the inflow rate increased. This occurs because a higher inflow rate speeds up the water’s advance phase and reduces drag force, leading to lower η values⁴⁸.

It should be noted that the cutoff time (T_co) is significantly influenced by irrigation events and inflow rates. In this study, the mean T_co were 52 min, 57.56 min, and 55.33 min for CFI, and 64.44 min, 74.56 min, and 71.11 min for AFI during the 1st, 3rd, and 5th irrigation events, respectively. As the flow rate increased, the T_co decreased. The highest T_co was observed during the 3rd irrigation event when surface resistance was at its maximum due to increased vegetation growth. These results align with the findings of Mazarei et al.⁶ and Mazarei et al.⁴⁸, which similarly demonstrated a direct correlation between cutoff times and Manning’s roughness. As Manning’s roughness increases, surface resistance also increases, leading to longer cutoff times.

Several researchers have noted that the temporal variation in infiltration parameters and Manning’s roughness significantly impact surface irrigation performance^19,45,48,55. Even with the same soil type, infiltration characteristics can vary depending on the irrigation method used^9,56, owing to differences in soil moisture content and the horizontal suction of surface soil under different treatment conditions.

Statistical evaluation of simulated advance and recession time

Advance time and recession time are key factors that significantly impact the efficiency and effectiveness of furrow irrigation systems. In this study, the advance time for CFI ranged from 23 to 68 min, shorter than that of AFI, where the advance time varied between 42 and 79 min. Similarly, the recession time ranged from 38 to 77 min for CFI and from 52 to 89 min for AFI (Fig. 6). A similar trend in advance and recession trajectories was reported by Yadeta et al.¹¹. Previous research has shown that factors like inflow rate, infiltration parameters, field geometry, and Manning’s roughness influence advance time^6,24,45. In this study, field geometry, and slope were kept consistent across all treatments, suggesting that any differences in advance times for the same furrow lengths can likely be attributed to variations in infiltration parameters and Manning’s roughness²⁴. Additionally, a noteworthy difference was observed in the advance times between the first irrigation (when furrows were newly formed) and subsequent irrigations in the same furrows. These differences were expected due to the influence of microrelief and surface roughness on the movement of shallow water⁴⁵. Notably, advance times were shorter at the upstream end of the furrows and gradually increased toward the downstream end throughout the irrigation process. This was largely due to the uneven distribution of soil moisture, where the upstream side of the furrow remained wetter while the downstream side remained drier. As a result, higher advance times and lower recession times were recorded at the downstream end, while the upstream end exhibited lower advance times and higher recession times¹⁰. These spatial variations highlight the dynamic nature of furrow irrigation and its dependence on infiltration and surface conditions.

Fig. 6

Measured and simulated advance and recession trajectories under CFI and AFI.

The simulation results for advance and recession times are summarized in Table 4. For CFI, the average RMSE, NS, and d values were 6.55 min, 0.956, and 0.859 for advance time, and 4.798 min, 0.953, and 0.904 for recession time. For AFI, the average RMSE, NS, and d values were 4.732 min, 0.904, 0.806, respectively for advance time and 3.202 min, 0.922, 0.877, respectively for recession time. For advance time across the events, RMSE, NSE, and d for CFI ranged from 3.64 to 13.7 min, 0.87 to 0.99, and 0.76 to 0.93, respectively. For AFI, the value varied from 0.61 to 7.22 min, 0.85 to 0.93, and 0.76 to 0.88. Increasing the inflow rate to 2 lps improved simulation accuracy for both irrigation methods. This improvement could be attributed to the increased water flow, which likely results in a more uniform and predictable advance trajectory, reducing the variability at lower inflow rates. Mazarie et al.⁶ and Araujo et al.⁵⁷ observed that inflow rates directly impact the advance trajectory, with lower rates showing greater sensitivity. Studies by Xu et al.⁴⁵ Chen et al.⁵⁸, and Sayari et al.⁵⁹, also indicated that WinSRFR provided accurate advance time simulations. The recession trajectories showed good alignment with measured values. For CFI, RMSE, NSE, and d values were between 1.59 and 11 min, 0.65 and 0.99, and 0.88 and 0.93, respectively. And for AFI, the RMSE, NSE, and d values were between 2.32 and 4.31 min, 0.87 and 0.95, and 0.79 and 0.93. The results show that the AFI method generally produced more accurate estimates than the CFI method, especially in recession times. This could be due to the AFI method’s adaptability to changing field conditions, allowing for more precise water flow and distribution control. Xu et al.⁴⁵ found that recession trajectories are influenced by the infiltration coefficient (k), which responds to water infiltration rate and duration changes. The simulation accuracy in estimating recession times was highest at a 2 lps inflow rate across all irrigation events, with slightly reduced accuracy at 1.5 lps compared to 1 lps and 2 lps. Mazarei et al.⁶ confirmed that WinSRFR adequately simulated both advance and recession trajectories but with reduced accuracy at lower inflow rates.

Table 4 Evaluation of measured vs. simulated recession trajectories using performance indicators.

Interestingly, the WinSRFR software performed better in simulating recession times than advance times for both irrigation methods, suggesting a potential bias in the software’s internal calculations or a limitation in capturing dynamic variables affecting advance time. The consistency of these results with previous studies^{54,56,59,60,61} reinforces WinSRFR’s applicability across various conditions and irrigation practices. However, it’s essential to recognize the limitations identified in this study, particularly at lower inflow rates where the software’s accuracy declined little. Addressing this issue might benefit from soil-specific calibration to help reduce these discrepancies and enhance the software’s accuracy in practical applications. Additionally, as suggested by Nie et al.⁶², incorporating spatial and temporal variations in soil infiltration rates could further refine the software’s estimates and improve its predictive performance.

To validate the reliability of the WinSRFR model in simulating furrow irrigation performance, paired t-tests and 95% confidence intervals were calculated for observed and simulated values of AE, DP and DU_lq across three different inflow rates (1 lps, 1.5 lps, and 2 lps). The results indicated that the mean differences were statistically insignificant (p > 0.05), suggesting good model agreement. Statistical analysis revealed that the model predictions for AE, DP and DU_lq were largely consistent with field measurements, particularly at 1 lps (Fig. 7). The paired t-tests showed no significant differences for AE (p = 0.397) and DP (p = 0.272) at 1 lps, with 95% confidence intervals including zero. However, at higher flow rates (1.5 lps and 2 lps), AE exhibited significant differences (p < 0.05), indicating potential model sensitivity to increased discharge rates. This phenomenon may be attributed to faster advance times and greater water movement dynamics at higher inflows, which can increase spatial variability in infiltration and complicate accurate simulation. Similar findings have been reported by Gillies et al.¹⁵, who observed that model accuracy tends to decrease at higher flow rates due to more considerable variability in furrow hydraulics. Mailapalli et al.¹⁶ also noted that higher discharge levels led to discrepancies in simulated AE, largely because of uneven infiltration patterns and runoff behavior. Furthermore, Koech et al.¹⁹ confirmed that higher inflow rates negatively impacted model precision for AE and DP, emphasizing the importance of careful calibration under variable operational conditions. For DP, the model showed mixed performance (Fig. 7). At 1 lps, a significant difference between observed and simulated values was detected (p = 0.012), with the 95% confidence not encompassing zero. This suggests that at lower inflow rates, the model may slightly underestimate or overestimate percolation losses, possibly due to variability in field infiltration characteristics or sensitivity to input parameter calibration. However, at higher flow rates of 1.5 lps and 2 lps, no significant differences were observed (p = 0.753 and p = 0.141, respectively), indicating improved model accuracy for DP at higher discharges. In terms of DU_lq, the model consistently performed well across all tested flow rates (Fig. 7). Statistical analysis showed no significant differences between observed and simulated DU_lq values (p = 0.111 at 1 lps, p = 0.471 at 1.5 lps, and p = 0.655 at 2 lps), and the confidence intervals consistently included zero. These findings demonstrate the model’s robustness in simulating water distribution patterns within the furrows.

Fig. 7

Observed vs. simulated values of Application Efficiency, Deep Percolation, Distribution Uniformity at 1 lps, 1.5 lps and 2 lps inflow rate for CFI and AFI methods. The solid black line represents the regression line, and the shaded region shows the 95% confidence band for model predictions.

Overall, the statistical evaluation confirms that the WinSRFR model provides reliable predictions for AE and DU_lq across varying operational scenarios. These results underline the necessity of fine-tuning simulation models like WinSRFR when applied across a range of discharge scenarios to maintain predictive reliability.

Simulation of irrigation performance indicator

The hydraulic performance indicators (AE, DU_lq, and DP) were assessed based on the assumption of a target water application equal to the calculated soil moisture deficit. As shown in Table 5, the mean observed AE values were 62.94%, 59.50%, and 58.94% for the 1st, 3rd, and 5th irrigation events, respectively. The mean simulated AE values were 61.88%, 54.77%, and 57.88% for the same events. The results suggest that WinSRFR tends to slightly underestimate AE. Additionally, AE decreased during the 3rd and 5th irrigation events compared to the 1st event, with average reductions of 5.46% and 6.35% observed under field conditions, and 11.48% and 6.46% under simulated conditions, respectively. This reduction in AE is likely due to higher cutoff times, increased surface roughness, and changes in the k infiltration parameter during the growing season^6,45. The table also indicates that increasing the inflow rate from 1 to 1.5 and 2 lps led to an average reduction in AE by 13.6% and 19.65%, respectively, for CFI, and by 12.8% and 13.7%, respectively, for AFI under field conditions and simulated conditions, reductions of 15.58% and 21.0%, respectively, for CFI, and 5.89% and 13.4%, respectively, were noted for AFI. Although higher inflow rates shortened the cutoff time, the increased infiltrated volume led to more deep percolation, causing over-irrigation. Additionally, longer furrow lengths generally resulted in higher AE. The highest AE values 66% for CFI and 85% for AFI were recorded with a furrow length of 65 m and an inflow rate of 1 lps under field conditions. Similarly, Holzapfel et al.⁶³ found that longer furrow lengths improve AE, while higher discharge rates and extended cutoff times reduce it. Under simulated conditions, the highest AE values (66% for CFI and 82% for AFI) were achieved with a 65-m furrow length and an inflow rate of 1 lps. This analysis identified this treatment as the optimal scenario in both the WinSRFR software and the field volume balance. Kang et al.⁶⁴ observed that infiltration was greater in CFI compared to AFI, resulting in higher AE for AFI. Lima et al.⁶⁵ corroborated that optimal furrow irrigation performance with continuous flow is achieved when the inflow rate is close to the minimum allowable.

Table 5 Results of measured and simulated performance indicators under the different treatments.

According to Table 5, the mean observed values of DU_lq were 83.89%, 84.00%, and 85.67% for the 1st, 3rd, and 5th irrigation events, respectively, while the corresponding simulated values were 83.55%, 82.88%, and 85.2%. No significant change was found between the 1st and 5th irrigation events; however, a noticeable decrease was observed in the 3rd event when comparing the simulated and observed values. The average DU_lq observed under field conditions was slightly higher than the simulation results. Zai et al.²⁴ and Nie et al.⁵³ highlighted that infiltration variability plays a significant role in influencing distribution uniformity and must be considered in practical applications. Furthermore, Araujo et al.⁵⁷ and Mazarei et al.⁶ demonstrated that factors such as field geometry (e.g., slope and length) and variable inflow parameters (e.g., inflow rate and cutoff time) have a considerable effect on DU_lq. DU_lq was also enhanced in AFI, reaching 91% under optimal parameters compared to 83% in CFI. This improvement is due to better control over water advance rates and cutoff timing in AFI, reducing the risk of over-irrigation at the inlet and under-irrigation at the tail end. These observations are in line with Qureshi et al.⁷, who highlighted the role of precise cutoff management in improving DU_lq in furrow systems. The results of this study showed that the largest DU_lq values were obtained during the 5th irrigation event, suggesting that DU_lq improves as the growing season progresses. The lower DU_lq observed during the 1st irrigation event may be attributed to higher infiltration rates, which also affected AE, as discussed earlier. Several researchers, including Hanson et al.⁶⁶ and, Lecina et al.⁶⁷ have observed DU_lq values exceeding 80%, which is consistent with the findings of this study. These results suggest that the irrigation system performed well in terms of uniform water distribution throughout the growing season.

The mean values of DP under field conditions were 37.06%, 40.50%, and 41.61% for the 1st, 3rd, and 5th irrigation events, respectively. In comparison, the simulated values were 38.44%, 41.55%, and 41.3% for the same irrigation events. A direct relationship between the number of irrigation events and DP was observed, with the DP values rising as the growing season progressed. Overall, the field-measured DP increased by 9.28% and 12.27% during the 3rd and 5th irrigation events compared to the 1st event. Similarly, the simulated DP values showed increases of 8.09% and 6.92% for the 3rd and 5th events, respectively, relative to the 1st event. It was also apparent that DP was higher in CFI compared to AFI, likely due to longer cutoff times and larger inflow volumes during the growing season. On average, DP was 34.5% higher in CFI compared to AFI, making AFI more suitable for regions with limited water availability. The higher DP in CFI is attributed to continuous water application to every furrow, leading to extended infiltration durations, especially at the furrow inlet zones. These findings correspond with the outcomes of Du et al.¹⁷, who reported higher DP losses in conventional furrow systems due to prolonged irrigation times. There was also a direct relationship between the inflow rate and DP; as the inflow rate increased from 1 to 1.5 lps and 2 lps, DP also increased. This was because the longer opportunity time at 2 lps led to the highest DP values⁶. The inefficiencies in water use, including deep percolation and poor distribution uniformity, were caused by prolonged advance and recession times, leading to greater infiltration near the furrow entrance and reduced infiltration at the downstream end⁴⁵. Our findings showed that increasing furrow length and reducing stream size led to a decrease in DP. This is because longer furrow lengths and smaller stream sizes prolonged the advance phase, thereby reducing water losses¹³. These findings align with research that highlights the importance of soil type and spatial variability in managing furrow irrigation^65,68. Mazarei et al.⁶, Nie et al.⁶², Salahou et al.⁵⁵, Chen et al.⁵⁸, and Sayari et al.⁵⁹ also confirmed the satisfactory accuracy of the WinSRFR software in simulating furrow irrigation performance, consistent with the findings of this study.

The comparative analysis of CFI and AFI revealed clear advantages of AFI in improving irrigation performance. Across the study, AFI consistently outperformed CFI in terms of Application Efficiency (AE) and Distribution Uniformity (DU_lq), particularly at moderate inflow rates (1.5 lps). These findings align with Kang et al.⁶⁴ who similarly observed enhanced water productivity and reduced deep percolation (DP) under AFI systems. AFI notably minimized DP losses, especially at lower flow rates, reducing water percolation beyond the root zone. This supports the observations of Du et al.¹⁷, who emphasized AFI’s role in improving water use efficiency through alternating wetting and drying cycles. These benefits are particularly valuable for water-scarce regions, enabling significant water conservation without yield penalties. Optimization of cut-off times and inflow rates further confirmed that under AFI at 1.5 lps, AE exceeded 75% and DUlq surpassed 85%, while keeping DP minimal. These findings provide practical guidance for farmers to fine-tune irrigation scheduling based on resource availability. Additionally, AFI offers operational benefits by reducing labor and energy costs, consistent with Chukalla et al.⁶⁹, who reported lower pumping requirements. Environmentally, AFI mitigates risks of soil erosion and nutrient leaching compared to CFI at high flow rates⁷⁰, supporting soil health preservation. Overall, this study provides empirical and model-based evidence promoting AFI as an effective water-saving strategy. The results support integrating tools like WinSRFR into decision-support systems to deliver tailored, efficient irrigation recommendations for diverse field conditions.

Optimization of furrow irrigation performance

Using WinSRFR, the optimal inflow rate and cutoff time were recommended for all furrow treatments, as shown in Table 6. For the 65-m furrow length, the optimum cutoff time was 64.4 min under a discharge of 1.1 lps, yielding the highest AE of 71% for CFI. For AFI, the highest optimized AE was 87%, achieved with a cutoff time of 80 min under the furrow length of 65 m and a discharge of 1 lps. For a furrow length of 55 m, the optimized AE reached up to 68% for CFI and 86% for AFI. This significant difference can be attributed to the reduced wetted surface area in AFI, which limits infiltration losses and allows for more uniform water distribution along the furrow length. These findings are consistent with Sarker et al.⁴¹, who demonstrated that AFI could improve water productivity by up to 25% compared to conventional methods. The observed superiority of AFI can also be linked to soil texture in the study area, where the clayey soil composition favors longer infiltration times. Notably, the 45-m furrow length showed the most significant improvement, with AE values of 65% for CFI and 80% for AFI after reducing the cutoff time by 5.88% to 22.78%, respectively. Furrow length of 65 m and 1 lps inflow rate yielded the highest optimized AE, ranging from 65 to 71% for CFI and 80% to 87% for AFI. This was followed by the 1.5 lps rate, which yielded AE values ranging from 57 to 62% for CFI and 68% to 73% for AFI, and the 2 lps rate, which produced AE values ranging from 55 to 57% for CFI and 62% to 71% for AFI. These results demonstrate that increases in inflow rate and cutoff time negatively impact AE, emphasizing the importance of optimizing these parameters to save water in furrow irrigation⁷¹. This trend is consistent with previous studies by Anwar et al.⁵⁴, and Mehri et al.⁴², which also found that increasing inflow rates and cutoff times reduce irrigation efficiency. Conversely, furrow length had the opposite effect, longer furrows resulted in higher Holzapfel et al.⁶³ and Xu et al.⁴⁵ also reported increased AE values with longer furrow lengths. These management combinations achieved more than 90% distribution uniformity, although AE varied across different inflow rates and cutoff times. The optimum cutoff time ranged from 30.30 min to 80 min. For the 45-m furrow length in CFI, the flow did not reach the downstream end below a cutoff time of 30.30 min, and for AFI, the minimum cutoff time was 40.40 min. Beyond a cutoff time of 64.4 min for the 65-m furrow length in CFI and 80 min for the same length in AFI, overflow conditions occurred. Research on optimizing inflow rates and cutoff times^{6,54,62,72,73} has concluded that effective management of these parameters can greatly enhance furrow irrigation performance, which aligns with the findings of this study.

Table 6 Results of the optimized value of inflow rate, cut-off time, application efficiency, and distribution efficiency.

Sensitivity analysis results

The sensitivity analysis results (Table 7) indicated that inflow rate exhibited the highest influence across all performance indicators, with sensitivity coefficients of 0.91 for AE, 0.85 for DP, and 0.72 for DU_lq, classifying it as a critical parameter in furrow irrigation modeling. Cutoff time showed moderate sensitivity, particularly affecting AE (0.58) and DP (0.65), confirming its importance in managing water application timing and reducing losses. Among infiltration parameters, the basic infiltration rate (f₀) and infiltration coefficient (k) displayed moderate sensitivity (0.44 and 0.35 for AE, respectively), underlining their relevance in controlling water infiltration dynamics. In contrast, Manning’s roughness coefficient (n) and furrow slope demonstrated low sensitivity, indicating minimal impact on performance metrics within the tested variation range. These findings align with previous research by Sayari et al.⁵⁹ and Ebrahimian et al.⁷⁴, who highlighted the dominant role of inflow rate and infiltration parameters in surface irrigation efficiency. Overall, the analysis confirms that accurate measurement and calibration of inflow rate and infiltration characteristics are essential for improving simulation reliability and guiding effective irrigation management. Mehri et al.⁴² also highlighted that inflowrate and cutoff times are most significant factors affecting application efficiency. The results of the sensitivity analysis in this study are consistent with previous similar studies^74,75.

Table 7 Sensitivity coefficient of some input parameters of the WinSRFR software.

External validation using optimized irrigation parameters

External validation of the WinSRFR model was conducted using independent field data collected during the subsequent irrigation season (2024–2025). The optimized irrigation parameters from the first season: namely discharge rate, cutoff time, and furrow geometry were applied under both Continuous Furrow Irrigation (CFI) and Alternate Furrow Irrigation (AFI) methods. Observed performance indicators, including Application Efficiency (AE), Deep Percolation (DP), and Distribution Uniformity (DU_lq), were compared with model-simulated values (Table 8). While slight deviations were observed between simulated and originally optimized values, these differences can be attributed to natural field variability and changes in environmental conditions, which influence infiltration dynamics and water distribution. Additionally, because WinSRFR simulations incorporate second-season conditions, the outputs may shift marginally even when applying fixed optimized parameters. These variations are expected in real field applications and emphasize the importance of external validation to ensure the practical reliability of model-based recommendations across seasons.

Table 8 External validation of the WinSRFR model using independent field data from the subsequent irrigation season.

Statistical evaluation confirmed the robustness of model predictions during external validation, with R² values of 0.97 for both AE and DP, and 0.93 for DU_lq, indicating strong correlation with field observations (Table 9). Low RMSE values and high correlation coefficients (r > 0.97) further supported the model’s predictive accuracy. These results validate that the optimized parameters derived from simulation can be effectively implemented to improve on-farm irrigation performance. The model’s performance aligns with the findings of Ahmadi et al.²³, who demonstrated comparable accuracy using WinSRFR under semi-arid conditions. Notably, unlike other studies, this research incorporates comprehensive field validation across multiple irrigation events, capturing temporal variability and enhancing result reliability an approach supported by Qureshi et al.⁷. These findings reinforce the practical utility of the WinSRFR model in guiding irrigation management decisions and offer actionable insights for improving water use efficiency, boosting crop productivity, and promoting sustainable furrow irrigation practices.

Table 9 Statistical evaluation of external validation analysis.

Limitations of the study

This study provides valuable insights into optimizing furrow irrigation methods; however, several limitations warrant consideration. The experiments were conducted under uniform soil conditions, which do not capture the natural variability in field conditions such as soil texture, infiltration capacity, and organic matter content. These factors can significantly influence water distribution and the accuracy of irrigation models. Although infiltration parameters were calibrated, local heterogeneities may not have been fully captured, potentially affecting model reliability. Field measurements including advance-recession times, soil moisture levels, and inflow rates are susceptible to human error and instrument limitations. Even minor inaccuracies in these data points can impact model calibration and, consequently, the precision of irrigation efficiency simulations. The WinSRFR model itself has inherent constraints. It assumes uniform furrow geometry and does not dynamically adjust for sediment deposition or erosion during irrigation events. This simplification can lead to inaccuracies in simulating real-world conditions. Moreover, the model relies heavily on precise input data; any inaccuracies can propagate through simulations. Additionally, it does not account for factors such as evaporation or wind effects, which can be significant in arid climates. The study focused exclusively on sunflower cultivation under specific climatic conditions, limiting the generalizability of the findings to other crops or environments with different evapotranspiration dynamics. Future research should explore multi-seasonal and multi-location trials, integrate automated data collection methods, and consider a broader range of environmental variables to enhance the robustness and applicability of the findings.

Practical implications and recommendations

The findings of this study provide actionable strategies for improving furrow irrigation management in various agricultural contexts. AFI showed better water use efficiency compared to CFI, especially at flow rates between 1.0 and 1.5 lps. AFI reached higher application efficiency and distribution uniformity while effectively reducing deep percolation losses. For practitioners, adopting AFI and modifying cutoff times based on specific field advance times can significantly decrease water losses below the root zone. Regularly recalibrating field parameters after significant weather events or field operations is advised to maintain optimal irrigation efficiency. Water resource managers can advocate for AFI as a water-saving strategy on a regional scale. Encouraging straightforward field monitoring practices, such as measuring advance times, empowers farmers to make timely and informed adjustments. ​ The relevance of these results extends across different soil types and climatic conditions. In sandy soils, AFI helps prevent rapid percolation losses, while in clay soils, managing flow rates is essential to avoid surface runoff. Incorporating these recommendations into policy frameworks and decision-support tools can encourage wider adoption. Mobile advisory applications linked to models like WinSRFR can offer real-time guidance for farmers, enhancing on-farm water management and contributing to sustainable agricultural practices.

Conclusions

This study highlights the practical utility of integrating simulation models with field experimentation to enhance surface irrigation efficiency in Indian agricultural settings. By calibrating and validating the WinSRFR model under real-world conditions, the research provides field-tested guidelines for improving furrow irrigation performance, particularly in clayey soils common across central India. The research confirms that furrow length and inflow rate are not merely operational variables, but strategic levers for minimizing deep percolation losses and improving water distribution uniformity. Using the WinSRFR model, both continuous and alternate furrow irrigation methods were optimized for maximum water application efficiency and minimal deep percolation losses. The alternate furrow irrigation method (AFI) achieved superior results, reaching an application efficiency of 87% with a flow rate of 1.0 lps and a cutoff time of 80 min, under a 65-m furrow length. This suggests that AFI is more effective in reducing water loss compared to continuous furrow irrigation (CFI), which achieved an application efficiency of 71% under optimized conditions. The results emphasize that irrigation strategies should adapt dynamically to crop growth stages, as both infiltration behavior and surface resistance change over time. The model successfully captured the infiltration dynamics and flow advance-recession behavior observed in the field. Time series outputs showed good agreement between observed and simulated advance times, indicating correct parameterization of infiltration coefficients and Manning’s roughness factors. These results validate the WinSRFR model’s capability to simulate surface irrigation hydraulics under variable field conditions, supporting its use as a decision-support tool for optimizing irrigation management. This combined experimental-simulation approach enhances confidence in the model outputs and provides region-specific guidelines that can be readily adopted by farmers and irrigation managers. Given the high sensitivity of inflow rate and cutoff time, real-time monitoring or automation of these parameters could further improve water use efficiency in future field applications.

Importantly, external field validation confirmed the accuracy and reliability of the optimized parameters, with high correlation between observed and simulated values and minimal error margins. Observed AE, DP, and DU_lq values closely matched simulated results, with validation RMSE values below 1.4% and correlation coefficients above 0.98. These results reinforce the practical applicability of the WinSRFR model for field-level irrigation planning. These findings provide clear, evidence-based recommendations for improving on-farm water management. Farmers and irrigation managers can adopt the optimized discharge rates and cutoff times identified in this study to enhance irrigation efficiency, reduce water wastage, and promote sustainable agricultural practices. The results are directly applicable to real-world field conditions and can support improved water resource management in similar agro-environments.