Journal of Researches in Mechanics of Agricultural Machinery

Abstract Abstract
In this study, the performance of a backpack sprayer equipped with a T-jet nozzle was evaluated under laboratory conditions. To investigate the amount of liquid droplet deposition on each pot, water-sensitive papers were used. The water-sensitive papers were then analyzed using image processing techniques with ImageJ software. This research was conducted based on a factorial experiment arranged in a completely randomized design, including the factors of forward speed at three levels (2, 3.5, and 5 km h⁻¹), spraying pressure at three levels (2, 3, and 4 bar), and distance of the pots from the fan at three levels (1, 2, and 3 m), with three replications. In all measurements, the nozzle height above the ground was maintained at 90 cm. The results showed that with increasing forward speed, the volumetric diameters of 10%, 50%, and 90% and the surface coverage decreased by approximately 11% and 31%, respectively. However, this decreasing trend was not statistically significant at all speed levels, such that no significant difference was observed between the speeds of 3.5 and 5 km h⁻¹ at the 5% probability level. The minimum and maximum solution deposition were obtained at pressures of 4 bar (19.37%) and 3 bar (28.53%), respectively. The highest surface coverage percentage was observed in pots located at the farthest distance from the fan. Ultimately, the highest surface coverage percentage was achieved in the treatment with a forward speed of 2 km h⁻¹ and a spraying pressure of 3 bar.
Introduction
In recent decades, the extensive use of pesticides and herbicides in agriculture has raised significant environmental and health concerns. Excessive and improper application can lead to adverse effects, including water contamination, biodiversity loss, and risks to human health. Consequently, enhancing the efficiency of pesticide application has become a critical priority for sustainable agriculture. One of the most effective approaches to improving pesticide application is to ensure uniform spray distribution and producing droplets of appropriate size, as droplet size directly influences spray coverage, drift, and pesticide effectiveness.
Previous research has shown that spraying performance is significantly affected by factors such as nozzle type, spray pressure, travel speed, and the physicochemical properties of the spray liquid. key indicators for evaluating spray quality include surface coverage percentage, volume median diameter, and spray uniformity, which are typically assessed using image analysis and water-sensitive papers. While many studies have addressed large-scale field spraying, relatively limited research has focused on potted plants under controlled environments such as greenhouses. This study investigates the influence of operational parameters, specifically forward speed, spray pressure, and distance of the fan from the pots on spray deposition characteristics in eggplant plants grown in pots. The goal is to evaluate droplet distribution and spray uniformity in relation to plant surface coverage. By applying cost-efficient image processing methods, the research aims to provide practical insights for improving pesticide application in greenhouses and small-scale farming systems.
Material and Methods
This study was conducted using a mobile-frame system at the Department of Biosystems Engineering, Shiraz University, to evaluate spraying performance under controlled conditions. A backpack sprayer (METEC SPRAYER) equipped with a TJ60-11004 nozzle was mounted on a galvanized steel frame designed for a 3 m spray width. The system was powered by a 1.1 kW electric motor (Motogen, Tabriz), and its speed was regulated using an inverter (Hyundai N50, South Korea). An axial fan generated a 3 m/s airflow parallel to the spray path, calibrated using an anemometer (Lutron YK-2001 AL). Eggplant seedlings were cultivated in pots, and water-sensitive papers were placed around plants to assess spray coverage and drift. After spraying, cards were scanned (HP Scanjet 3770, 2400 dpi) and analyzed with ImageJ 1.46r. A factorial design in a completely randomized layout was employed, with three levels of pressure (2, 3, 4 bar), speed (2, 3.5, 5 km/h), and fan distance from the pots (1, 2, 3 m). Data were analyzed using Duncan’s test at the 5% significance level with SPSS (version 26).
Results and Discussion
The results of the variance analysis indicated that forward speed had a significant effect on the volumetric diameter parameters (DV0.1, DV0.5, and DV0.9) and surface coverage percentage at the 5% probability level. Additionally, spray pressure, fan distance from the pots, and the interaction between forward speed and pressure on surface coverage demonstrated statistically significant differences. Increasing the forward speed from 2 to 5 km/h led to a decrease of approximately 11% in volumetric droplet diameters, attributed to reduced droplet deposition on target surfaces due to shorter residence time. Surface coverage also decreased by nearly 31% as forward speed increased, due to the nozzle's shorter exposure time over the plants. At forward speeds of 2, 3.5, and 5 km/h, the average surface coverage was 29.07%, 23.94%, and 20.06%, respectively. Spray pressure significantly influenced surface coverage. The highest surface coverage of 28.53% was observed at 3 bar pressure, while the lowest coverage of 19.37% was observed at 4 bar pressure. The distance of the fan from the pots affected spray deposition; pots located farther from the fan received greater coverage due to droplet drift. No significant difference was observed between those pots positioned closer to the fan. The interaction between spray pressure and forward speed was significant at the 5% probability level. The highest surface coverage was observed at 3 bar and 2 km/h, while the lowest was recorded at 4 bar and 5 km/h. Overall, higher forward speeds and pressures led to smaller droplets and greater drift, reducing overall spray deposition efficiency on target surfaces.
Conclusions
This study investigated the spray performance of a backpack sprayer equipped with a TeeJet nozzle under varying operational conditions. The results demonstrated that forward speed and spray pressure significantly affect both droplet size and surface coverage. Specifically, increased forward speed led to a reduction in volumetric droplet diameters (DV0.1, DV0.5, DV0.9) and a notable decline in surface coverage, primarily due to reduced nozzle residence time over the target area. Among the evaluated spray pressures, 3 bar provided better overall coverage than 2 and 4 bar, likely due to lower droplet drift at lower pressures. Fan distance also played a critical role in deposition patterns, with pots farther from the fan receiving greater coverage due to the redistribution of airborne droplets. Optimal spraying performance was achieved at a spray pressure of 3 bar, forward speed of 2 km/h, and fan distance of 3 m. These findings offer practical insights for improving pesticide application efficiency in greenhouse or semi-controlled environments. Further research is recommended to validate these results under field conditions and to explore the influence of additional environmental factors, such as wind direction, humidity, and canopy structure, on spray dynamics.
Acknowledgements
The authors gratefully acknowledge Shiraz University for its financial support.
Author Contributions
Author1: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Conceptualization.
Author2: Visualization, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Conceptualization.
Author3: Visualization, Validation, Software, Methodology, Investigation, Formal analysis, Data curation, Conceptualization.
Author4: Writing – review & editing, Visualization, Validation, Software, Methodology, Investigation, Formal analysis. Author5: Writing – review & editing, Visualization, Validation, Software, Methodology, Investigation, Formal analysis. Author6: Writing – review & editing, Visualization, Validation, Investigation. Author7: Writing – review & editing, Visualization, Validation.
Data Availability Statement
Data will be made available on request.
Ethical Considerations
The authors confirm that research ethics were considered throughout the research methodology, including data definition, data collection, data analysis, data interpretation, and plagiarism, to provide the final report.
Conflict of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper
Funding Statement
This work was supported by grant number 148321 from Shiraz University.

Abstract Abstract
In this research, a device for peeling green almond skins with the ability to use single-phase electricity. This study aims to design and evaluate an almond peeling machine. The mechanical design of the machine components was carried out based on design principles Three variables including the rotational speed of the drum, its distance from the concave, and the thickness of the rubbers were selected as the test factors. Statistical analysis showed that the rotary speed of the crusher and its distance from the concave significantly affected the peeling efficiency (with a confidence level of 1%). Taguchi analysis indicated that the ranks of 1 to 3 were obtained for the factors of impactor speed, distance between impactor and concave, and rubber thickness on the capacity of the device respectively. In relation to comparing the effect of the factors of the speed of the threshing machine, the distance between the threshing machine and the concave and the thickness of the rubbers on the percentage of almond breakage by the device, shows the ranks 1 to 3 for these three factors, respectively. The device was successfully designed and built and its performance was evaluated based on practical needs and application expectations. The results showed that the designed device was able to achieve the intended goals well.

Introduction
Post-harvest processing of almonds involves removing the outer green skin (hull) from the almonds, which is an important step in product preparation. The moisture content of the green skin has a significant impact on the ease and quality of peeling. Almond kernel breakage is considered as one of the key indicators in the process of evaluating product quality. Cracking or breaking of the hard shell of almonds can cause mechanical damage such as breakage or scratches on the kernel, which ultimately leads to a decrease in the marketable quality of the product. The design and development of an almond peeling machine is considered an important technical need, given the increasing need for mechanization in agricultural processes. This machine reduces production costs by improving the speed and accuracy of the peeling process. Green almond shellers are mainly designed based on the principles of controlled abrasion and impact. The most common design of these machines consists of a shelling cylinder in which the almonds are guided into a horizontal lattice or cage cylinder. This study aims to design and evaluate an almond peeling machine. This device is designed to be able to operate with single-phase and three-phase electricity, combustion engines, and tractor power, making it easy for farmers and gardeners to use. It has the ability to separate the green skin of almonds from their hard skin, and to clean all types of stone and paper almonds.
Material and Methods
In this research, a device for peeling green almond skins with the ability to use single-phase electricity, a combustion engine, or a tractor's power axle was designed, manufactured, and evaluated. The main components of the machine include the chassis, feed hopper, thresher, concave, carrying handles, tractor attachment arms, bearings, and carrier wheels. Three variables including the rotational speed of the drum, its distance from the concave, and the thickness of the rubbers were selected as the test factors. In evaluating the performance of the device, the effect of the parameters of the agitator rotational speed at three levels (150, 300, and 540 rpm), the distance of the agitator from the grooved cylinder at three levels (20, 30, and 50 mm), and the thickness of the rubber at three levels. ‌Surface (60, 80, and 100 mm) was evaluated in a factorial design based on randomized complete blocks on peeling efficiency and fracture percentage of almond kernels.
Results and Discussion
The mechanical design of the machine components was carried out based on design principles and was manufactured and assembled taking into account a sufficient safety factor in each part of the machine. Based on the results of variance analysis, the factors of the rotary speed of the crusher, the distance between the crusher and the concave, and the thickness of the rubbers had a direct effect on the peeling efficiency and the percentage of broken hard skin of almonds. Statistical analysis showed that the rotary speed of the crusher and its distance from the concave significantly affected the peeling efficiency (with a confidence level of 1%). The percentage of broken skin was also affected by the distance between the crusher and the counter-crusher, the rotary speed, and the thickness of the rubbers. In terms of comparing the effects of the three factors considered in the present study, Taguchi analysis indicated that the ranks of 1 to 3 were obtained for the factors of impactor speed, distance between impactor and concave, and rubber thickness on the capacity of the device, and the delta values for these three factors were 277.8, 188.8, and 37.5, respectively. In terms of optimal levels for increasing efficiency, based on the results obtained, the highest speed level, the lowest distance level, and the middle level of rubber thickness were seen to be better. Also derived from the same analysis, the increasing trend of the capacity of the device has occurred with increasing the speed of the threshing and conversely with decreasing the distance. A similar study has been conducted in relation to comparing the effect of the factors of the speed of the threshing machine, the distance between the threshing machine and the concave and the thickness of the rubbers on the percentage of almond breakage by the device, which shows the ranks 1 to 3 for these three factors, respectively, with delta values of 0.389, 0.232 and 0.106. Also, the lowest level of threshing speed, the highest level of distance and the middle level of the thickness of the rubbers, have included a better signal-to-noise ratio.
Conclusions
The device studied in this research was successfully designed and built. Considering the existing research needs, its performance was evaluated based on practical needs and application expectations. The results showed that the designed device was able to achieve the intended goals well.The findings show that in the design and construction of almond shellers, the quality of the threshing surfaces, the speed, and the blade spacing are key factors in determining efficiency. The decrease in efficiency with increasing distance between the thresher and the counter-thresher indicates the need for a balance between efficiency and durability, which can be considered in future designs. Some of the advantages of the manufactured device include: the ability to operate the device with different power sources, the ability to effectively separate green almond skins completely and uniformly, the ability to clean hard and papery almonds with minimal, significant economic savings such that the total price of this device, compact design and easy transportation (which makes it easy to move and use in different conditions), the ability to adapt to peeling other nut products, and the availability of after-sales service and technical support in the country due to the localization of the production process.

Acknowledgements
The authors would like to acknowledge the Urmia University for finantcial support.

Author Contributions
Ghafari Azad: Methodology
Mardani Aref:Supervisor
Sharifian Faroogh: Supervisor
Data Availability Statement
Not applicable.
Ethical Considerations
Authors proclaim that ethical considerations have been observed in conducting this study and confirm adherence to ethical standards, including avoidance of data fabrication, falsification, plagiarism, and misconduct.
Conflict of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Conflict of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper
Funding Statement
The authors received no specific funding for this research.

Abstract Abstract
Apple fruit is subjected to operations such as harvesting, packaging, grading, storage, and transportation. During these stages, numerous static and dynamic loads are applied to individual fruits, leading to mechanical damage, quality deterioration, and consequently a reduction in marketability and export potential. The objective of this study was to analyze the dynamic behavior of export apples of the Golden Delicious and Red Delicious cultivars against impact-induced damage using finite element simulation. In addition, using an Instron texture analyzer available in the rheology laboratory, the mechanical properties of the apple peel, flesh, and core were determined separately. The measured physical and mechanical properties were then used as input data in the engineering data section of Abaqus software. To achieve the objectives of the study, an impact test was first conducted in a separate section as a factorial experiment with independent variables including two apple cultivars, three impact energy levels (0.0124, 0.0238, and 0.0712 J), and two curvature radii of the impact location (38.8 and 42.2 mm). The results of this study indicated that, in the simulation of the apple impact test using the finite element method, there was a high correlation and reliability between experimental results and modal analysis, such that the coefficient of determination (R²) obtained from the comparison of experimental and modal frequencies was 0.98 and 0.99 for Golden Delicious and Red Delicious apples, respectively. The results of this study can provide a significant advantage in the design of harvesting and post-harvesting machinery for apples at a commercial scale.

Introduction
Apples are affected by operations such as harvesting, packaging, grading, storage, and transportation on their way from the orchard to the sales centers. During these stages, various static and dynamic loads are applied to each product, which leads to damage and also causes a decrease in quality, and as a result, a decrease in marketability and export of the product. This research aims to analyze the dynamic behavior of export apples of Golden Delicious and Red Delicious varieties against damage caused by impact using finite element simulation, which is defined by defining a viscoelastic material model in the two parts of the flesh and core (central part) of the apple and an elastic-plastic material model in the apple skin section using finite element modal analysis using Abaqus software. In order to investigate the linear regression model of the dynamic behavior of apple fruit, an impact was applied to the apple fruit using three-axis force gauge and accelerometer sensors. The physical properties of the apple fruit were measured in the laboratory. Also, using the Instron texture analyzer available in the rheology laboratory, the mechanical properties of the skin, flesh, and seed of apple fruit were determined separately, and the results of the physical and mechanical properties were used as input in the engineering data section of the Abaqus software. To achieve the objectives of the study, first, in a separate section, an impact test was performed in the form of a factorial experiment with independent variables including two apple varieties, three levels of impact energy (0.0124, 0.0238, and 0.0712 Joules) and two levels of the radius of curvature of the impact site (38.8 and 42.2 mm). The results of this study showed that in simulating the apple impact test using the finite element method, there is a high correlation and reliability between the experimental results and modal analysis, so that the coefficient of determination (R2) resulting from the comparison of experimental frequencies and modal analysis for the Golden Delicious and Red Delicious apple varieties is 0.98 and 0.99, respectively. So that the results of this study can provide a significant advantage for the design of harvesting and post-harvest machines on a commercial scale for apple fruit.
Material and Methods
To conduct this research, healthy, undamaged export apples of the Golden Delicious and Red Delicious varieties were selected from orchards in Mahabad city (longitude 34.5°45'47" and latitude 9.7°51'36"), the hub of apple production and export in the country, and were harvested manually. The harvested product was transferred to the rheology laboratory of the Faculty of Agriculture, Bu Ali Sina University, in special containers under standard conditions. The fruits were then stored in a refrigerator at 4 degrees Celsius. Before conducting the experiments, the apple fruits were placed at the laboratory temperature to become isothermal with the ambient temperature. Then, mechanical tests and impact tests were performed on the samples.
Results and Discussion
To determine the natural frequencies of apple fruit by impact testing, the output data from the excited samples were recorded by a four-channel signal processing device (ECON, AVANT Lite, model: MI-6004), and the results were then transferred from the time domain to the frequency domain using the fast Fourier transform in MATLAB software. In the polar direction, the first vibration mode for the Golden Delicious cultivar was sensed at a natural frequency of 125.27 Hz, and the second and third modes had natural frequencies of 151.12 and 185.17 Hz, respectively (Table 4). In the Red Delicious cultivar, the first vibration mode was sensed at a natural frequency of 121.61 Hz, and the second and third modes had natural frequencies of 148.21 and 174.23 Hz, respectively, in the polar direction (Table 6). The first modal frequency (simulated) in the polar direction in the Golden Delicious cultivar was considered to be 138.18 Hz (first mode), the second modal frequency was considered to be 151.45 Hz (second mode), and the third modal frequency was considered to be 204.14 Hz (third mode) (Table 4). Also, in the Red Delicious variety, the first modal frequency in the polar direction was 132.18 Hz (first mode), the second was 155.61 Hz (second mode), and the third was 197.30 Hz (third mode) in the Abaqus simulation (Table 6). In the equatorial direction, the first vibration mode for the Golden Delicious variety was sensed at a natural frequency of 135.52 Hz, and the second and third modes had natural frequencies of 141.16 Hz and 175.17 Hz, respectively (Table 5). In the Red Delicious variety, the first vibration mode was sensed at a natural frequency of 124.18 Hz, and the second and third modes had natural frequencies of 158.25 Hz and 214.23 Hz, respectively, in the equatorial direction (Table 7). The first modal frequency in the equatorial direction in the Golden Delicious cultivar was 141.38 Hz (first mode), the second modal frequency was 155.86 Hz (second mode), and the third modal frequency was 187.45 Hz (third mode) (Table 5). For the Red Delicious cultivar, the first modal frequency in the equatorial direction was 137.16 Hz (first mode), the second modal frequency was 168.42 Hz (second mode), and the third modal frequency was 227.22 Hz (third mode) in the Abaqus software simulation (Table 7). As shown, the differences between these frequencies for both apple fruit groups are highly reliable, so that the linear regression model and the coefficient of determination R2 resulting from the comparison between the experimental and modal frequencies for the Golden Delicious and Red Delicious varieties are 0.98 and 0.99, respectively. These results can provide a significant advantage in the design of commercial-scale harvesting and post-harvest machines for apple fruit.
Conclusions
In the industry, before manufacturing engineering parts, modal tests are used to determine the material, selecting a more compatible material for parts with greater displacement to extend their life and improve function. Considering this type of application of modal analysis, this approach can be used to genetically modify agricultural products, especially fruits, to change their texture, thereby increasing their useful life. Also, the finite element model is a fast, cost-effective, and non-destructive method for evaluating natural frequencies, especially the resonant frequency, and for providing a regression model with acceptable accuracy for the design of harvesting and post-harvest fruit machines.
Acknowledgements
Bu-Ali Sina University is appreciated for supporting this research in terms of supplying laboratory materials and equipment.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Ethical Considerations
Conflict of Interest: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper or any conflict of interest.
Ethical Review: This study does not involve any human or animal testing. Informed Consent: Not require
Conflict of Interest
There are no contributions of interest in this article, and this has been confirmed by all authors.
Funding Statement
The author(s) received no specific funding for this research.

Abstract Abstract
Visual recognition of disease symptoms in plant leaves involves observing changes in leaf appearance that can indicate disease or plant stress. These symptoms include spots, discoloration, wilting, leaf rust, abnormal growth, and abnormal leaf drop. These symptoms can be identified and managed using advanced visual tools, such as remote sensing and drone imagery, and artificial intelligence-based algorithms. Therefore, present paper aimed to identify plant diseases in different crops using image processing and deep learning algorithms. The models including RegNet, ShuffleNet, and DenseNet were evaluated based on criteria such as accuracy, precision, recall, F1 score, and ROC and accuracy-recall curves. The overall accuracy of the RegNet model is about 96%, indicating it successfully identifies most rice and wheat diseases. The overall accuracy of the ShuffleNet model was 94.4%, a promising result, though slightly lower than that of the previous model. The highest accuracy, 97.2%, was achieved with the DenseNet model.
Introduction
The life cycle of agricultural products undergoes significant changes due to diseases. Visual recognition of disease symptoms in plant leaves involves observing changes in leaf appearance that can indicate disease or plant stress. These symptoms include spots, discoloration, wilting, leaf rust, abnormal growth, and abnormal leaf drop. These symptoms can be identified and managed using advanced visual tools, such as remote sensing and drone imagery, and artificial intelligence-based algorithms.
Meshram et al. (2021) reviewed the applications of machine learning algorithms in agriculture in three phases: pre-harvest, during-harvest, and post-harvest. This review found that machine learning algorithms have achieved significant results in solving agricultural problems. Their study showed that there is a need to follow the path of machine learning with standard empirical methods. Researchers should create their own datasets and make them available on different platforms so others can use them to test and validate their models. Rahmn et al. (2025) developed a real-time leaf disease detection system for 8 plant species (potato, tomato, bell pepper, apple, corn, grape, peach, and rice) using deep learning techniques, including Inception, VGG, MobileNet, and DenseNet. Their dataset consisted of 30,945 images and 35 disease classes. The highest accuracy for disease detection was 98% for tomatoes with the Inception model, 100% for bell peppers with the MobileNet model, 100% for apples with the MobileNet model, 99% for grapes with the VGG model, 100% for peaches with the VGG model, and 98% for rice with the DenseNet model.

Material and Methods
A total of 13,324 images of diseased leaves from 5 different crops, namely, cotton, potato, sugarcane, wheat, and rice (Figure 1), with different image sizes but a total of 4.79 GB, were obtained from the Kaggle site. Table 1 shows the setup and implementation features of the proposed model training.
Proposed Models
1. ShuffleNet Model
The ShuffleNet model is a convolutional neural network architecture with a lightweight computational structure and is less complex than models such as VGG and ResNet. Its disadvantages include its possible lower performance compared to models such as EfficientNet.
2. RegNet Model
The RegNet model is another convolutional neural network architecture that is highly flexible for computer vision applications such as image classification, object recognition, and segmentation. Compared to complex architectures such as VGG, the network is simple and efficient, and offers a better balance between accuracy and computational cost.
3- DenseNet Model
Unlike standard architectures that only connect between consecutive layers, the DenseNet model connects all layers together. This means that because each layer adds new learned features to subsequent layers, the number of parameters is reduced, and the model consumes less memory.
To evaluate the efficiency of the classifiers, the criteria of precision, accuracy, recall, and F-score were extracted from the confusion matrices, the operating characteristic curve diagram, and the precision-recall diagram.
Results and Discussion
The ResNet model does not overfit; both the training and validation datasets have good accuracy, and the training and validation errors have decreased and then stabilized. This means the model has been able to train on new data effectively. In the ShuffleNet classifier, it was observed that the training accuracy increased rapidly. The validation accuracy also increased but did not reach the training accuracy, indicating mild overfit. The DenseNet model also showed an increase in accuracy, but validation accuracy fluctuated and did not reach the desired value. This model was not stable in learning new data.
In all models, most values lie on the main diagonal of the matrices, indicating that the models make correct predictions across classes. The model performs well across most classes but may be biased in some.
In all three proposed models, the lines for many classes are close to the upper right corner, indicating high precision and recall. This means that the model performs very well in recognizing most classes. Classes whose lines are not close to the upper corner may need improvement, which may be due to class similarity or data imbalance.
The overall accuracy of the RegNet model is about 96%, indicating it successfully identifies most rice and wheat diseases. The overall accuracy of the ShuffleNet model was 94.4%, a promising result, though slightly lower than that of the previous model. The highest accuracy, 97.2%, was achieved with the DenseNet model.
Conclusions
This study aimed to identify plant diseases in different crops using image processing and deep learning algorithms. The results showed that different models, including RegNet, ShuffleNet, and DenseNet, performed well at classifying plant diseases. The models were evaluated based on criteria such as accuracy, precision, recall, F1 score, and ROC and accuracy-recall curves.
• The RegNet model performed stably and showed the best balance between accuracy and recall.
• The ShuffleNet model had a slight overfit, but overall provided acceptable performance.
• The DenseNet model, despite its dense connections, had fluctuations in validation accuracy in some cases and needs improvement.
• The ROC and accuracy-recall curves showed that the models performed very well in identifying most classes. Still, some classes need improvement, which may be due to feature similarity or data imbalance.
Finally, the results of this research confirm that the use of deep learning algorithms in plant disease identification can effectively increase agricultural productivity and be an important step towards smart agriculture.
Author Contributions
Sajad Sabzi: Conceptualization
Razyieh Pourdarbani: Writing
Mahan Masoumzade: Methodology
Data Availability Statement
"Not applicable".
Ethical Considerations
It is confirmed adherence to ethical standards, including avoidance of data fabrication, falsification, plagiarism, and misconduct.
Conflict of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper
Funding Statement
The author(s) received no specific funding for this research

Abstract Abstract
Sour cherry juice, prized for its high antioxidant, phenolic, and natural pigment content, is a popular beverage that is highly susceptible to microbial spoilage. To assess microbial inactivation, a continuous-flow system was constructed comprising a sour cherry juice reservoir, a high-precision pump, an argon gas tank for plasma formation, and a custom-designed glass Venturi tube reactor. Plasma was generated in the liquid phase using injected argon gas under voltages ranging from 10 to 20 kV.Sour cherry juice was autoclaved at 121°C for 15 minutes to eliminate background microorganisms and then inoculated with E. coli cultured for 24 hours. Treatment parameters included voltage, flow rate (2–8 L/min), temperature (30–50°C), and time (1–5 min). Under optimized conditions voltage of 19.85 kV, flow rate of 3.52 L/min, temperature of 47.82°C, and treatment time of 4.42 min the hybrid system achieved a maximum E. coli reduction of 6.18 log cycles (99%). In addition to microbial inactivation, the hybrid system showed superior retention of phenolic compounds, anthocyanins, vitamin C, and color integrity, which are key quality markers in fruit juice processing. Energy demand remains a limiting factor, necessitating integration with energy-saving systems and optimization of plasma operating conditions.
Introduction
Sour cherry juice, prized for its high antioxidant, phenolic, and natural pigment content, is a popular beverage that is highly susceptible to microbial spoilage. This vulnerability presents significant safety and shelf-life challenges for producers. Conventional thermal pasteurization, while effective in microbial inactivation, often degrades the nutritional and sensory quality of the product. As consumer demand grows for minimally processed beverages with preserved nutritional and functional characteristics, non-thermal processing methods have become increasingly important. Among them, liquid-phase plasma and hydrodynamic cavitation stand out due to their high microbial inactivation potential without substantial heat generation. However, few studies have examined their synergistic effects in fruit juice processing.
This study addresses this gap by designing and optimizing a novel hybrid system that combines a Venturi tube-based hydrodynamic cavitation reactor with liquid-phase plasma technology. The aim is to inactivate Escherichia coli (E. coli) in sour cherry juice while minimizing the degradation of heat-sensitive compounds. The core innovation lies in the concurrent use of physical (cavitation) and chemical (reactive species) stress mechanisms. By leveraging the strengths of both methods, the system aims to deliver high microbial inactivation with minimal impact on physicochemical quality. This study also investigates the relationships among voltage, flow rate, temperature, and treatment time in determining system effectiveness using Response Surface Methodology (RSM). The results have the potential to advance sustainable, non-thermal pasteurization technologies in the beverage industry.
Material and Methods
To assess microbial inactivation, a continuous-flow system was constructed comprising a sour cherry juice reservoir, a high-precision pump, an argon gas tank for plasma formation, and a custom-designed glass Venturi tube reactor. The reactor featured a throat diameter of 3.14 mm, length of 17.69 mm, and convergence/divergence angles of 23° and 80.81°, respectively. Tungsten electrodes were used due to their high melting point (>3000°C). Plasma was generated in the liquid phase using injected argon gas under voltages ranging from 10 to 20 kV.
Sour cherry juice was autoclaved at 121°C for 15 minutes to eliminate background microorganisms and then inoculated with E. coli cultured for 24 hours. Treatment parameters included voltage, flow rate (2–8 L/min), temperature (30–50°C), and time (1–5 min). Microbial load was determined using the plate count method on MacConkey agar. Experiments were conducted in triplicate. RSM with a Box-Behnken design was employed for optimization.
Results and Discussion
Under optimized conditions voltage of 19.85 kV, flow rate of 3.52 L/min, temperature of 47.82°C, and treatment time of 4.42 min the hybrid system achieved a maximum E. coli reduction of 6.18 log cycles (99%). The regression model showed excellent predictive power with an R² value of 0.99, standard error of 0.059, and coefficient of variation of 2.10%. ANOVA revealed that treatment time and voltage were the most significant variables (P < 0.1), followed by temperature and flow rate. Significant interactions were found between several variables, and quadratic effects highlighted complex interdependencies.
Quality analyses compared untreated, thermally treated, and hybrid-treated samples. Total phenolic content (TPC) in untreated juice was 283.71 mg GAE/100 g, decreasing by 16.5% (to 237.11 mg GAE/100 g) with thermal treatment and 14.2% (to 243.51 mg GAE/100 g) with hybrid treatment. Total anthocyanin content (TAC) dropped from 241.23 mg C3G/L to 201.57 mg C3G/L (thermal) and 217.32 mg C3G/L (hybrid), indicating better pigment retention with the non-thermal method.
Vitamin C content showed a significant difference: 41.94 mg/L in untreated samples declined to 19.62 mg/L (thermal, 53.2% loss) and 29.67 mg/L (hybrid, 29.3% loss). Color preservation, measured via spectrophotometry, indicated a lower color difference (ΔE = 2.52) for the hybrid-treated juice compared to thermal treatment (ΔE = 3.86), preserving the natural red hue more effectively.
The combination of reactive oxygen and nitrogen species from plasma (•OH, ozone, peroxynitrite) and extreme pressure and temperature from cavitation synergistically enhanced microbial inactivation. The physical damage from bubble collapse complemented chemical attacks on cell membranes and DNA, improving penetration and efficacy.
Conclusions
This study demonstrates that integrating liquid-phase plasma and Venturi-induced hydrodynamic cavitation offers a powerful, non-thermal method for inactivating E. coli in sour cherry juice. The approach ensures microbial safety while preserving critical nutritional and sensory properties, outperforming traditional thermal treatments. The high accuracy of the RSM model validates its utility in predicting and optimizing treatment parameters for maximal efficacy.
In addition to microbial inactivation, the hybrid system showed superior retention of phenolic compounds, anthocyanins, vitamin C, and color integrity, which are key quality markers in fruit juice processing. Its scalability, however, requires further consideration. Energy demand remains a limiting factor, necessitating integration with energy-saving systems and optimization of plasma operating conditions.
Future work should focus on industrial-scale implementation, cost-benefit analysis, and application of this technology to other beverages with different physicochemical characteristics and microbial loads. Ultimately, this dual non-thermal technology represents a sustainable and efficient alternative for producing high-quality, microbiologically safe fruit juices, with strong potential for commercial adoption.
Author Contributions
The data supporting the findings of this study are available from the corresponding author upon reasonable request.
Ethical Considerations
This study did not involve any human or animal participants. All microbiological procedures involving Escherichia coli were conducted under appropriate biosafety level 1 (BSL-1) conditions. The authors affirm that the study complies with ethical standards and no data fabrication, falsification, plagiarism, or research misconduct occurred during the research process.
Conflict of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper
Funding Statement
This work was supported by Research and Technology Council of Shahrekord University .

Abstract Abstract
The neural network technique is a powerful modeling method that mimics biological memory and learning using neurons. In this study, the classification of packaging types under dynamic and quasi-static loading conditions on the banana fruit properties were investigated. The input data for the neural network included the phenolic content, antioxidant capacity, acidity, pH, soluble solids, firmness, and moisture content. The classification output data consisted of the different packaging types were used in the experiment. A perceptron network with 3 and 5 neurons in different layers was employed, utilizing the hyperbolic tangent as the selected activation function. In this research, the performance of multilayer artificial neural networks for classifying and predicting experimental data were evaluated. The models were trained with varying numbers of learning epochs and iterations, and the results were assessed based on mean squared error (MSE) values. The best performance was achieved with the MSE of approximately 0.00019 for the training data and 0.02 for the validation data, indicating the high accuracy of the model under optimal conditions. Additionally, increasing the number of learning epochs and iterations generally led to reduced error and improved prediction accuracy. However, in some cases, extremely low training error values suggested the possibility of model overfitting.

Introduction
More than one - third of agricultural crops are wasted after harvest that leads to lack of food. Improving storage methods, especially in developing countries, can reduce waste, increase food security and support farmers. The mechanical properties of the material are more comprehensive and clear understanding of the physical properties and consequently the texture. Employing this knowledge is useful to reduce the damage during the required operation and to increase the shelf life of fruits . Hence, attention to the shape and physical dimensions are required to be used for separation and sorting. some of physical properties such as shape, size, surface area, density, porosity, color, and appearance are important in many problems depending on design of special machines or analyzing product behavior in material handling. Also, the appearance characteristics of fruits are affected by their value in the market. Hence, it is important to properly control the harvest and post-harvest operations. In recent years, several studies have been conducted to perform a precise evaluation of food products due to increasing consumer demand and their special attention to the internal quality of fruit such as freshness, sweet and nutritional value. There are various researchers conducted by differenet researchers. Azadbakht et al, (2019) conducted a research on the classification of chemical properties of pears under dynamic loading using artificial neural network .the results showed that radial basis function network had the best performance in the classification of loads .fu et al. (2016) conducted a study to classify kiwifruit on the basis of shape using a image processing. They measured length , maximum diameter of basal area and projected area of fruit. The stepwise linear regression was used to select important variables in predicting the minimum diameter of basal area and volume. Hosnavi et al (2011) investigated physical-chemical properties to classify date palm fruits in 14 native cultivars from different areas of morocco , tunisia and algeria .the results showed that sugar content are the dominant part of date fruitss .Also , the amount of protein , fat and ash were observed in the samples .
The aim of present paper is classification of banana fruit by using its constituent components in static and dynamic loads using MLP neural network.
Material and Methods
Based on the opinion of experts from local markets in Gorgan, a sufficient number of semi-ripe bananas were procured. First, the defective samples were separated. Then, the bananas were sorted by size to make them uniform in size and weight, thus eliminating the fruits from the standard size. Then, the fruits were packaged and stored for 14 days. The chemical properties of the samples were measured daily.
Results and Discussion
The models were performed in 2 period of learning and 2 replications for performance analysis. the overall results showed that increasing the number of learning period caused error reduction and performance improvement, especially in class 2 which recorded the highest accuracy
In contrast, Class 3 and Class 4 still had high errors, indicating the need for further optimization.
Also , the R value was negative, suggesting a mismatch between model predictions and actual data .the increase in the number of model runs had a significant effect on accuracy improvement in some classes .For further optimization, it is suggested to use methods such as weights, selecting the optimal architecture of neural network and optimization algorithms such as Adam or RMS prop.
Conclusions
The results of the implementation of neural networks showed that the accuracy of the model is strongly influenced by the number of learning periods and the number of replications. The increase in the number of replications, reduced the mean square error (MSE). In the case of the number of learning cycles of 1000, the final error in cross-validation is higher, indicating the possibility of overfitting in low-complexity models.
On the other hand, implementing learning period of 2000 resulted in a significant decrease in educational error and also improved model performance in experimental data .in some cases , very low MSE values indicate that the model accurately learns the training data , but there is a need to validate the performance of the model over the new data to ensure that the model is appropriate .Adjusting the weights , reducing learning rate, and increasing the number of training data can help improve model performance and prevent the overfitting.
Author Contributions
Razyieh Pourdarbani: Conceptualization
Sajad Jafarzadeh: Writing
Farid Moradi: Methodology
Data Availability Statement
"Not applicable".
Ethical Considerations
It is confirmed adherence to ethical standards, including avoidance of data fabrication, falsification, plagiarism, and misconduct.
Conflict of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper
Funding Statement
The author(s) received no specific funding for this research

Abstract Abstract
The aim of this research is the development of hybrid models based on transfer learning concept to classify apple images into four classes named: healthy apples, apples affected by bacterial disease, apples affected by fungal disease, and apples with tissue level disease propagation. A single-staged four-class model as well as six hybrid models were utilized for classification. Each hybrid model consisted of a three-class model followed by a two-class model. To train them, 850 images of the Kaggle dataset was used. The backbone of all transfer learning models was the EfficientNet algorithm. The models were implemented in the PyCharm environment, using the Python programming language. Evaluation of the models were carried out using the confusion matrix, as well as the calculation of the precision, accuracy, sensitivity, specificity, and F1-score criteria. According to the results of this study, the value of all performance evaluation criteria exceeded 0.94, indicating that the developed models performed robustly in classifying apple images; however, some hybrid models performed worse than the single-staged classifier. Therefore, the optimal configuration of hybrid classifiers cannot be assumed a priori; instead a systematic evaluation of all possible combinations is necessary to identify the most effective approach.

Introduction
Recent advances in precision agriculture highlight the pivotal role of machine vision and deep learning in developing automated systems for fruit classification and grading. A wide range of studies have been conducted with the primary aim of replacing manual methods and improving accuracy and efficiency. These studies can be broadly categorized into three main approaches.
The first involves hybrid frameworks, where fast detection models such as YOLO are integrated with advanced classification architectures, achieving accuracies of approximately 92% across multiple fruit types. The second focuses on multi-stage systems, designed for specific fruits, in which the classification process is structured into sequential steps such as fruit type identification, freshness assessment, and ripeness evaluation. The third approach emphasizes explainable models, exemplified by frameworks such as XAI-FruitNet, which not only achieve very high accuracies (exceeding 97%) across diverse datasets but also provide interpretability in model decision-making. Overall, while these studies share a common objective—enhancing accuracy and overcoming the limitations of conventional approaches—they differ considerably in their technical methodologies and scope of application. Some prioritize feature integration, others employ multi-stage analysis, and yet others focus on explainability.
Building on these advancements, the present study proposes the use of sequential hybrid classifiers based on transfer learning to categorize apple images into four distinct classes: healthy apples, apples affected by bacterial disease, apples affected by fungal disease, and apples with tissue-level disease progression.
Material and Methods
The dataset used in this study was obtained from the Kaggle repository and consists of four categories of apple images:
• Bacterial Disease (Blotch): 234 images, divided into 146 for training and 88 for testing.
• Healthy Apples (Normal): 222 images, divided into 130 for training and 92 for testing.
• Advanced Tissue Rot (Rot): 226 images, divided into 134 for training and 92 for testing.
• Fungal Disease (Scab): 168 images, divided into 100 for training and 68 for testing.
The objective of this study was to classify apples into one of these four categories. Two classification strategies were investigated: a single unified four-class classifier and sequential hybrid classifiers. The hybrid classifier first classifies images into three classes, with one class combining two of the four original categories. Subsequently, a binary classifier further separates the images within this composite class into the two original groups.
Given the four-class problem, six possible hybrid configurations were constructed. The performance of each hybrid classifier was evaluated and compared with the unified four-class classifier. Using the abbreviations B, N, R, and S to represent the four categories (Blotch, Normal, Rot, and Scab, respectively), the classifiers investigated were as follows:
Unified four-class classifier: B–N–R–S
• Hybrid classifier BN–R–S followed by a binary classifier B–N
• Hybrid classifier BR–N–S followed by a binary classifier B–R
• Hybrid classifier BS–N–R followed by a binary classifier B–S
• Hybrid classifier B–NR–S followed by a binary classifier N–R
• Hybrid classifier B–NS–R followed by a binary classifier N–S
• Hybrid classifier B–N–RS followed by a binary classifier R–S
All classifiers employed the pre-trained EfficientNet model as the backbone, and the implementations were carried out in Python using the PyCharm environment. Performance evaluation was conducted using confusion matrices and standard metrics including accuracy, precision, sensitivity, specificity, and F1-score.
Results and Discussion
By employing the proposed model instead of a conventional convolutional neural network, the number of trainable parameters was drastically reduced. This substantial reduction not only decreased the training time of the transfer learning-based model but also enabled its deployment on computers without GPU support.
All performance evaluation metrics exceeded 0.94, indicating that the developed model performs robustly in classifying apple images. These results confirm the suitability of the transfer learning approach using the pre-trained EfficientNet architecture for categorizing apples into four classes: Bacterial Disease (Blotch), Healthy (Normal), Advanced Tissue Rot (Rot), and Fungal Disease (Scab).
Regarding hybrid classifiers, the B–NR–S and BS–N–R configurations outperformed the unified four-class classifier, while the BR–N–S hybrid achieved comparable accuracy. In contrast, the BN–R–S, B–NS–R, and B–N–RS hybrid classifiers performed worse than the unified model. This variation can be explained by the fact that combining classes acts like a form of feature engineering, and since all possible hybrid combinations are considered, this feature engineering occurs stochastically. Consequently, some combinations may lead to reduced overall accuracy. Therefore, when developing hybrid classifiers, it is essential to account for this property and discard configurations that underperform relative to the unified classifier.
Conclusions
This study demonstrates that all developed models achieved robust, reliable performance, with all evaluation metrics exceeding 0.94, which can be attributed to the use of a sufficiently large image dataset and the incorporation of transfer learning algorithms. The consistent high performance across models underscores the effectiveness of transfer learning in enhancing the accuracy and generalizability of automated apple classification systems.
A comparative analysis revealed that some hybrid models outperformed the unified four-class classifier, while others performed comparably worse. These findings indicate that the optimal configuration of hybrid classifiers cannot be assumed a priori; instead, a comprehensive development and systematic evaluation of all possible hybrid combinations are necessary, with each being benchmarked against the unified model to identify the most effective approach.
The relative advantage of the unified classifier stems primarily from its single-model architecture, in contrast to the paired structure of hybrid models. While some hybrid configurations offer potential improvements in classification accuracy, their deployment in an operational apple grading line requires additional hardware, which may pose practical challenges. This highlights a critical trade-off between achieving marginal performance gains and maintaining the feasibility of real-world implementation.
Overall, the findings of this study not only confirm the efficacy of transfer learning-based approaches for multi-class fruit classification but also provide important insights into the design considerations, performance trade-offs, and practical constraints associated with implementing hybrid versus unified classifier architectures in industrial agricultural settings.
Data Availability Statement
Data are available on request from the author.
Ethical Considerations
The author has utilized the Microsoft Copilot to summarize articles for writing the introduction section of the paper, and to translate Persian to English in writing the English abstract.
Conflict of Interest
The author declare that he has no known competing financial interest or personal relationship that could have appeared to influence the work reported in this paper.
Funding Statement
The author received no specific funding for this research.

Abstract Abstract
conservation tillage has attracted considerable attention due to its role in preventing erosion, maintaining soil organic matter, preserving soil moisture, and reducing water loss. This study aimed to compare different conservation tillage methods under periodical irrigation conditions with a conventional tillage method under continuous irrigation conditions (control treatment) for rice cultivation over two years (2020–2021). The experiment was conducted using a randomized complete block design with three replications at the Shavoor Station of the Khuzestan Agricultural Research Center. The treatments included four conservation tillage methods: no-till, chisel with disc, chisel packer, once disc, under periodical irrigation conditions, and conventional method: (control treatment: moldboard plow with twice disc under continuous irrigation conditions). The results of agronomic traits showed that there was no significant difference between different tillage-irrigation methods in terms of yield and agronomic traits, except for thousand-grain weight. Comparison of means showed that, no-till, chisel with disc, chisel packer, once disc, under periodical irrigation conditions, had 3137, 2900, 2892, and 2940 kg.ha-1 respectively, and control treatment had 3509 kg.ha-1 in yield. In terms of seed weight, the four conservation tillage methods under periodical irrigation conditions were at the same level and between 19.3 and 19.6 grams, but the control treatment achieved the highest amount with 20.6 grams. Also, the results showed that there was a significant difference among different tillage-irrigation methods in terms of applied water and water productivity. Comparison of the means showed that the control treatment with 41084 and no-tillage- periodical irrigation with 22753 m3.ha-1 had the highest and lowest applied water, respectively. Also, in terms of water productivity, the no-tillage periodical irrigation with 0.128 and control treatment with 0.084 kg.m-3 obtained the highest and lowest values, respectively.
Introduction
Conservation tillage is one of the important foundations of sustainable agriculture. The transition from conventional tillage to conservation tillage is underway in many developing countries. The advantages of conservation tillage compared to conventional tillage, depending on different conditions, can include reduced energy consumption, reduced water and wind erosion, less labor required, increased soil organic matter, and reduced operation time and increased operational efficiency. Also, a more important issue is reducing water consumption and increasing its productivity in agriculture, in which conservation tillage can play a role. Drought and climate change have made water issues a top priority in all fields, so conservation tillage on the one hand, and water-reduction patterns on the other, are being developed, taking into account the advantages of each. Therefore, in order to protect the soil and increase its productivity in reduced-tillage or no-tillage systems and also to reduce water consumption in periodical irrigation methods, considering recent droughts and climate change conditions, such a study was conducted. This study aimed to compare different conservation tillage methods under periodical irrigation and control conditions (conventional tillage- continuous irrigation method used by most farmers).
Material and Methods
This study aimed to compare different conservation tillage methods under periodical irrigation conditions, and the conventional tillage method under continuous irrigation conditions (control treatment) of rice, for two years (2020-2021) in a randomized complete block design with three replications, at the Shavoor Station of the Khuzestan Agricultural Research Center. The treatments included four conservation tillage methods: no-till, chisel with disc, chisel packer, once disc, under periodical irrigation conditions, and conventional method: (Control treatment: moldboard plow with twice disc under continuous irrigation conditions). The measured indicators included yield and yield components, including seed germination percentage, plant height, number of clusters per square meter, number of grains per cluster, harvest index, thousand grain weight, and biological yield. Also, economic indicators such as costs and incomes and two important indicators of irrigation water, including applied water and water productivity, were measured and evaluated. After completing the measurements and performing analysis of variance (ANOVA) using MSTATC software, comparison of the means of the indicators was performed using Duncan's multiple range test.
Results and Discussion
The combined analysis of variance indicated no significant differences among the tillage–irrigation methods between different tillage-irrigation methods in terms of yield and agronomic traits, except for thousand-grain weight. Also, the interaction effect of these methods with year did not have a significant difference in yield or any of the traits. Comparison of means showed that among different tillage- irrigation methods, no-till, chisel with disc, chisel packer, once disc, under periodical irrigation conditions, had 3137, 2900, 2892, and 2940 kg.ha-1 respectively, and control treatment (moldboard plow with twice disc under continuous irrigation conditions) had 3509 kg.ha-1 in yield. In terms of seed weight, the four conservation tillage methods under periodical irrigation conditions were at the same level and between 19.3 and 19.6 grams, but the conventional tillage method under continuous irrigation conditions achieved the highest amount, with about 5 percent more and with 20.6 grams. In terms of biological yield, the control was relatively higher than the other methods with 11026 kg.ha-1, and the lowest biological yield with 9936 kg.ha-1 belonged to the chisel packer- periodical irrigation. Also, the results showed that there was a significant difference among different tillage-irrigation methods in terms of applied water and water productivity, but no significant difference was observed in the interaction between tillage-irrigation methods and year in these two indices. Comparison of the means showed that the control treatment (moldboard plow with twice disc under continuous irrigation conditions) with 41084 and no-tillage- periodical irrigation with 22753 m3.ha-1 had the highest and lowest applied water, respectively. Also, in terms of water productivity, the no-tillage periodical irrigation with 0.128 and control treatment with 0.084 kg.m-3 obtained the highest and lowest values, respectively. In terms of cost, the control treatment had the highest cost, and the no-till-periodical irrigation method had the lowest cost with a rate of 0.91 compared to the control. Also, the net income assessment showed that all conservation tillage-periodical irrigation methods were less than the control, and among these methods, the no-till-periodical irrigation method had the highest net income with a rate of 0.89 compared to the control.
Conclusions
Based on the results of this study, in terms of crop yield, there is no significant difference between the different tillage-irrigation methods (four conservation tillage-periodical irrigation methods and conventional tillage-continuous irrigation method) in rice cultivation in Khuzestan Province, with the control treatment (conventional tillage with continuous irrigation) producing the highest yield, and the chisel packer-periodical irrigation obtained the lowest yield. In terms of applied water and water productivity, there is a significant difference between different tillage-irrigation methods in rice cultivation in Khuzestan Province, and the control (conventional tillage-continuous irrigation) achieved the highest applied water and lowest water productivity, while the no-till-periodical irrigation method achieved the lowest applied water and highest water productivity. Comparison between conservation tillage - periodical irrigation methods indicates better results of the no-tillage method in terms of higher yield, lower applied water, and higher water productivity than other reduced tillage methods. Therefore, for the development of rice cultivation in Khuzestan Province, considering its lower applied water and higher water productivity, conservation tillage methods such as no-tillage, under periodical irrigation, are recommended.
Acknowledgements
We would like to thank the Agricultural Research, Education and Extension Organization for its support in conducting this research.
Author Contributions
First Author: Conducting research in field conditions, collecting data, analyzing data, and writing a research report.
Second Author: Selecting treatments, determining the research implementation method, and consulting on research implementation.
Third Author: Consulting on how to measure water consumption indicators and monitor data collection and analysis of this data.
Data Availability Statement
Information and results are presented in the text of the article. Further details and required data will be provided upon request.
Ethical Considerations
The authors have observed ethical principles in conducting and publishing this scientific work, and this is confirmed by all of them. Also, in this study, artificial intelligence was not used in any of the research and writing of the article.
Conflict of Interest
The authors declare that there are no known competing financial interests or personal relationships that may have influenced the work reported in this article.
Funding Statement
This work was supported by the Agricultural Research, Education and Extension Organization under project number 0-14-14-011-990173.