Development of a Real-Time Irrigation Strategy Based on Cumulative Reference Evapotranspiration (ET₀) for Cabbage Cultivation in Paddy-Converted Fields

Abstract

This study developed an efficient cultivation strategy for cabbage production in paddy fields. To address poor drainage, discarded coir substrates (CS) were reused and compared with conventional paddy soil (PS). Four irrigation levels (ETc140, ETc100, ETc60, and ETc0) were applied to both CS and PS to evaluate their interactive effects. An automated irrigation system was deployed, integrating a weather sensor and solenoid valves via a LoRa-based IoT network. Hourly ET₀ was calculated based on Penman–Monteith in real time, and an irrigation event was triggered when cumulative ET₀ reached 1 mm (CS) or 3 mm (PS). The automated irrigation system showed stable performance. Hourly ET₀ estimates were 97% consistent with Korea Meteorological Administration data. The actual total irrigation depth (ID_actual) remained within 2% of the calculated depth (ID). Under moderate irrigation depths (ETc60 and ETc100), the reuse of CS significantly improved cabbage photosynthetic efficiency. Both CS-ETc60 and CS-ETc100 treatments maintained superior yield performance compared with other treatments. This integrated strategy not only offers a practical solution for improving water use efficiency but also enhances the multifunctional utilization of paddy fields, supporting the transition toward more sustainable agricultural practices.

1. Introduction

With the rapid development of greenhouse horticulture, soilless cultivation systems have been widely adopted. However, these systems generate large amounts of waste materials such as rockwool, peat, coir substrates, and plant residues [1,2,3]. Coir substrates come from natural, biodegradable sources and are increasingly regarded as a sustainable substitute for peat, with usage steadily rising [4]. Currently, most of these waste substrates are disposed of through landfilling or incineration [5]. However, landfilling can lead to soil and groundwater contamination, whereas incineration results in increased greenhouse gas emissions. Previous studies have demonstrated the potential for reusing coir substrates after hydroponic cultivation [6,7,8]. However, most research has focused on recycling within hydroponic systems, with limited exploration of their application in open field cultivation. Based on this, we have previously conducted studies on the reuse of waste coir substrates for upland cultivation in paddy fields, involving crops such as soybean and cabbage [9,10]. When converting paddy fields to upland cropping, poor drainage and prolonged waterlogging often cause soil hypoxia. This inhibits root respiration and severely restricts crop growth, yield, and quality [11,12]. The superior drainage and aeration characteristics of coir substrate can effectively alleviate root hypoxia and improve crop productivity and quality. Although coir substrates improve aeration and drainage, their unique physical properties complicate water management. Therefore, integrating substrate cultivation with precise and efficient irrigation technologies is critical to sustaining crop productivity. Given the increasing scarcity of water resources worldwide, enhancing crop water use efficiency is critical to sustainable agricultural development. Precision irrigation technologies are pivotal in achieving this goal [13].

Precise irrigation management depends on the accurate quantification of crop water requirements. To address this, a variety of modeling approaches have been developed. Among the most widely adopted are the FAO56 crop coefficient method and crop simulation models, both applied extensively in open-field [14,15,16]. The FAO56 method (Kc–ET₀) is particularly valued for its clarity, standardized parameters, and ease of use [17,18]. Recent studies have confirmed the robustness of FAO-provided Kc values across diverse crops and regions. Even under changing climate conditions, the ETc/ET₀ ratio has generally remained stable, with no evident systematic bias across most crops and regions, supporting the method’s reliability and broad applicability [19,20]. Alternatively, crop simulation models such as DSSAT and AquaCrop simulate crop growth, physiological processes, and soil–water dynamics to assess the impacts of different management practices on yield and resource use efficiency [21,22]. These models provide detailed insights into plant physiology and soil water dynamics but require more complex structures and extensive input data. As such, they are well-suited for scenario analysis and the development of decision support systems. By contrast, the FAO56 method remains more practical for irrigation scheduling in practical field settings, particularly where data are limited or real-time decisions are required [23,24].

As crop water requirement estimation methods continue to advance, automated irrigation systems integrating sensor technologies, the Internet of Things (IoT), and artificial intelligence algorithms have become a major focus in the development of precision irrigation [25]. These systems typically incorporate soil moisture sensors, weather monitoring instruments, and crop models to collect real-time data on environmental conditions and crop physiological status. By setting irrigation thresholds or running decision algorithms, the system autonomously regulates irrigation timing and volume, enabling dynamic management of crop water supply [26,27,28,29]. However, despite their capability for real-time root-zone moisture monitoring, soil moisture sensors face significant practical challenges. Factors such as installation accuracy, calibration status, sensor depth, and soil salinity can affect measurement accuracy [30,31]. Advanced sensors such as Time Domain Reflectometry (TDR) and Frequency Domain Reflectometry (FDR) measure soil dielectric constants and electrical impedance, respectively, but require complex installation and maintenance [32]. Furthermore, relying on just a few single-point sensors is insufficient to capture the spatial variability in soil moisture across large fields, which limits the reliability and representativeness of data. Ultrasonic sensors offer the advantage of large-area monitoring but are generally cost-prohibitive [33]. Accordingly, an automated irrigation system was designed and developed in this study based on the real-time estimation of crop reference evapotranspiration (ET₀), using cumulative ET₀ thresholds to regulate crop water supply dynamically. To test the potential of this approach for improving irrigation efficiency and conserving water resources, system performance, water use efficiency, and crop productivity were evaluated. Compared with irrigation strategies that rely on soil moisture sensors, the proposed system lowers equipment dependency and maintenance costs. It also enables timely and effective water management for large-scale crop production. By addressing the limitations of traditional soil moisture sensor-based strategies, the proposed system seeks to offer a more cost-effective and scalable solution for open field irrigation.

2. Materials and Methods

2.1. Experimental Site and Plant Material

The experiment was conducted at Chungnam National University in Daejeon, South Korea (36°22′ N, 127°21′ E, 45 m) from 16 April to 11 June 2024 (Figure 1a). The experimental site is characterized by a temperate monsoon climate, with an annual precipitation of approximately 1458.7 mm. Around 50% to 60% of the total precipitation occurs between late June and early September [34]. The experimental soil texture is classified as sandy loam according to the USDA Taxonomy (

Table 1. Physical properties of paddy soil (PS) and discarded coir substrates (CS).

Treatments	PD (g·cm−3)	BD (g·cm−3)	TP (%)	Sand (%)	Silt (%)	Clay (%)
PS (0~20 cm)	2.23	1.24	44.4	66.0	28.0	6.0
PS (20~40 cm)	2.32	1.25	46.1	68.0	26.0	6.0
PS (40~60 cm)	2.34	1.22	47.9	68.0	28.0	4.0
CS	1.36	0.07	94.9	–	–	–

PD—particle density; BD—bulk density; TP—total porosity.

). The soil chemical characteristics and profile show that the organic matter content below 20 cm is lower than in the topsoil (

Table 2. Chemical properties of paddy soil (PS) and coir substrates (CS).

Treatments	pH (1:5)	EC (dS·m−1)	Total N (%)	OM (g·kg−1)	AP (mg·kg−1)	Exch. Cations (cmolc·kg−1)
						K+	Ca2+	Mg2+
PS (0~20 cm)	6.2	0.3	0.11	17	188	0.36	3.4	0.7
PS (20~40 cm)	6.7	0.2	0.03	4	50	0.23	3.8	0.8
PS (40~60 cm)	6.9	0.2	0.01	1	11	0.18	2.9	0.8
CS	5.4	3.8	1.63	809.6	2594.13	1.04	75.03	10.64

Total N—total nitrogen; OM—organic matter; AP—available P₂O₅.

). A black and gray–green soil layer is present between 20 and 40 cm, and the soil below this becomes whitish (Figure 1b). Based on these characteristics, it is speculated that the soil may belong to the Arenosol type [35].

The four-leaf stage seedlings of ‘KoKoMa F1’ cabbage (Brassica oleracea var. capitata L.) were transplanted in the field on April 16. Each ridge measured 20 m² in area and 40 cm in height, with the surface covered by mulch film. A two-row planting pattern was used on each ridge, with an inter-row spacing of 60 cm and an intra-row spacing of 30 cm, resulting in approximately 133 plants per ridge. The trial adopted a split-plot design with three replicate ridges for each treatment. To eliminate the irrigation effects between plots, plastic boards were buried during ridge formation to isolate the plots.

2.2. Experimental Design and Cultivation Conditions

To evaluate the feasibility of reusing discarded coir substrate in paddy-converted fields, the main plot included two growth substrates: (1) Paddy soil (PS) treatment, consisting of paddy soil that was only plowed and fertilized after rice harvest, without drainage pipe installation; (2) Coir substrate (CS) treatment, consisting of discarded CS slab (100 × 20 × 10 cm; L × W × H, Daeyoung GS Co., Ltd., Daegu, Republic of Korea) that had been used for one year in hydroponic bell pepper cultivation. The substrates were then recycled and applied to the ridges for planting. The detailed cultivation method and treatment of the discarded coir substrate were described in our previous study [10]. The experimental field soil had been used for two years in upland crop planting experiments with discarded coir substrates, with only tillage and base fertilizer applied (Figure 1c,d). In order to apply the base fertilizer appropriately, paddy soil and discarded coir substrate samples were collected before the experiment and sent to the Daejeon Agricultural Technology Center for physical and chemical property analysis (Table 1 and Table 2). Based on the fertilizer formula provided by the Daejeon Agricultural Technology Center, a base fertilizer containing 129 kg∙ha⁻¹ of nitrogen (N), 304 kg∙ha⁻¹ of phosphorus (P₂O₅), and 114 kg∙ha⁻¹ of potassium (K₂O) was applied. Pest and disease management during the trial followed the cultivation standards provided by Asia Seed Co., Ltd., Seoul, Republic of Korea [36].

The subplot included four irrigation levels: over-irrigation (ETc140, 140% ETc), adequate irrigation (ETc100, 100% ETc), deficit irrigation (ETc60, 60% ETc), and no irrigation (ETc0, 0% ETc). The entire experiment utilized a drip irrigation system, with a drip emitter flow rate of 2 L·h⁻¹ (SEOWON Co., Ltd., Goesan-gun, Republic of Korea). During the first week after transplanting cabbage, an average of 500 mL of water was applied per plant daily until root establishment. The irrigation depth was calculated based on the product of the hourly ET₀ values and the crop coefficient (Kc) provided by FAO 56 [37], according to Equation (1).

ETc = ET₀ × Kc

(1)

ET₀ represents the reference evapotranspiration (mm∙h⁻¹), and Kc is the crop coefficient. Since all treatments were covered with plastic mulch, the adjusted Kc values for different crop growth stages are presented in

Table 3. The Kc values, cumulative ET₀, and cumulative ETc for each growth stage of cabbage, with 100% ETc as the baseline.

Growth Stage (Date Range)	kc	Cumulative ET0 (mm)	Cumulative ETc (mm)
root establishment (4.16–4.22)	–	–	–
initial (4.23–5.2)	0.56	34.19	19.15
development (5.3–5.22)	0.84	77.19	64.84
late (5.23–6.11)	0.76	85.83	65.23

.

$${\mathrm{ET}}_0={\displaystyle \frac{0.408\mathrm{\Delta }({\mathrm{R}}_{\mathrm{n}}-\mathrm{G})+\gamma \frac{37}{{\mathrm{T}}_{\mathrm{hr}}+273}{{\mathrm{u}}_2(\mathrm{e}}^0\left({\mathrm{T}}_{\mathrm{hr}}\right)-{\mathrm{e}}_{\mathrm{a}})}{\mathrm{\Delta }+\gamma (1+0.34{\mathrm{u}}_2)}}$$

(2)

where ET₀ is reference evapotranspiration (mm∙h⁻¹); R_n is the net radiation at the grass surface (MJ∙m⁻²∙h⁻¹); G is soil heat flux density (MJ∙m⁻²∙h⁻¹); T_hr represents hourly air temperature (°C); ∆ is saturation slope vapor pressure curve at T_hr (kPa∙°C⁻¹); γ is psychrometric constant (kPa∙°C⁻¹); e°(T_hr) is saturation vapor pressure at air temperature T_hr (kPa); e_a is average hourly actual vapor pressure (kPa); and u₂ is average hourly wind speed (m∙s⁻¹). The hourly net radiation, temperature, relative humidity, air pressure, and wind speed were calculated based on data collected by on-site sensors, as described in detail in Section 2.3 and Section 2.4.

The total irrigation depth (ID) was calculated for each treatment according to Equation (3).

$$\mathrm{ID}({\mathrm{m}}^3)={\displaystyle \frac{\mathrm{E}\mathrm{T}\mathrm{c}\times \mathrm{A}}{1000}}$$

(3)

where ETc is the crop evapotranspiration (mm), A is the area of one plot (20 m²), and 1000 is the conversion factor from mm to m³. The IDs of each treatment are presented in

Table 4. The calculated total irrigation depth (ID) for each treatment during the experiment.

Irrigation Level	ETc140	ETc100	ETc60	ETc0
ID (m3)	4.65	3.45	2.26	0.47

.

2.3. Automated Irrigation Algorithm

The irrigation algorithm used in this study proposed an automated control method based on cumulative ET₀. First, the weather sensor collected field environmental data every 3 min and used the hourly average data to calculate the hourly ET₀. Subsequently, the system accumulated the hourly ET₀ values and determined whether they exceeded a preset threshold. If the accumulated ET₀ reached the set threshold, the solenoid valve opening duration was calculated, and a command was sent through the LoRa gateway to trigger irrigation. During the irrigation process, the irrigation records were uploaded to the database to support data analysis and system optimization.

The irrigation threshold for cumulative ET₀ was set based on the different characteristics of the coir substrate (CS) and paddy soil (PS). Due to the high porosity and excellent drainage properties of coir substrate, irrigation at appropriate intervals can enhance crop water use efficiency. Therefore, the irrigation threshold for the CS treatment was set when the cumulative ET₀ reached 1 mm. In contrast, frequent irrigation in paddy soil with poor drainage can exacerbate surface waterlogging, which negatively impacts root health. To prevent this, and to achieve optimal soil moisture, the irrigation threshold for paddy field soil was set when the cumulative ET₀ reaches 3 mm (Figure 2).

2.4. Irrigation System Design

Figure 3 illustrates the overall architecture of the irrigation system. The system consists of an IoT gateway, edge devices, irrigation components (including solenoid valves, flow meters, and a pump), and a weather sensor. These units are interconnected via LoRa wireless communication technology, enabling data acquisition, transmission, and coordinated system control. The IoT gateway serves as the data hub of the system. It is responsible for receiving environmental parameters monitored by the weather sensor (WS500-UMB; Lufft, Fellbach, Germany) and the operational status of the irrigation solenoid valves (HPW2110A; HYOSHIN MECHATRONICS Co., Ltd., Bucheon-si, Republic of Korea) installed on-site. These data are then transmitted to the edge devices for processing. The edge devices are responsible for real-time processing of irrigation decisions based on the integrated automatic irrigation algorithm. All system units are powered by solar panels (KF20M-12; Kf solar Co., Ltd., Nanjing, China), and energy is stored in batteries (ATLASBX BXKB4.5–12; Hankook & Company Co., Ltd., Seongnam-si, Republic of Korea), ensuring stable operation of the system in the absence of an external power supply. This irrigation system can improve water use efficiency and reduce water waste by adopting precise irrigation with frequent, small irrigation amounts.

2.5. Irrigation System Evaluation

As the automated irrigation algorithm developed in this study uses on-site meteorological data, the accuracy of weather sensor readings is critical for reliable ET₀ estimation and irrigation scheduling. To verify the sensor’s reliability, hourly meteorological data were compared with corresponding records from the Korea Meteorological Administration (KMA, Daejeon Regional Office of Meteorology, 36°37.199′ N, 127°37.21′ E, 67.79 m) over the five days preceding the experiment (11–15 April 2024). Detailed meteorological data recorded during the entire experiment are presented in Table S1.

In addition, the accuracy of the automated irrigation system was assessed by comparing the ID (Table 4) with the actual total irrigation depth (IDactual) recorded by the flow meters. The flow meters (KTC 1-11-10; Korea Aichi Co., Ltd., Gimpo-si, Gyeonggi-do, Republic of Korea) recorded the irrigation depth for each treatment based on the monitored ETc values at the end of the experiment. No flow meter was installed for the ETc0 treatment, as irrigation was only applied during the root establishment period. Mechanical flow meters were used for monitoring, and their technical specifications and accuracy details are provided in

Table 5. The flow rate and accuracy specifications of the flow meters.

Parameter	Meaning	Flow Rate (m3·h−1)	Accuracy
Q3	Nominal flow rate	1.0	±2.0%
Q2	Transition flow rate	0.04	±2.0%
Q1	Minimum flow rate	0.025	±5.0%

. The actual flow rate for each treatment was 0.266 m³·h⁻¹ (calculated based on a drip emitter flow rate of 2 L·h⁻¹ and 133 plants per plot), which falls within the Q2–Q3 range, with a measurement accuracy of ±2.0%. The relative error was calculated according to Equation (4).

$$\mathrm{R}\mathrm{e}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e} \mathrm{e}\mathrm{r}\mathrm{r}\mathrm{o}\mathrm{r}=\left|\right(\mathrm{I}\mathrm{D}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{u}\mathrm{a}\mathrm{l}-\mathrm{I}\mathrm{D})/\mathrm{I}\mathrm{D}|\times 100\mathrm{\%}$$

(4)

2.6. Soil Moisture Monitoring

Monitoring of soil moisture dynamics provides an effective means to evaluate irrigation system performance and assess the impact of varying irrigation amounts on the root-zone moisture for cabbage cultivation. Therefore, TDR soil moisture sensors (MTE 7; Transducer System Electronics Co., Ltd., Seongnam-si, Gyeonggi-do, Republic of Korea) were installed in each treatment plot to monitor the volumetric water content (VWC) in the root-zone throughout the experiment. In the CS treatment, sensors were vertically inserted into the substrate following the standard installation protocol and positioned directly beneath the drip emitter at a depth of approximately 10 cm, based on the size of the coir substrate. To ensure consistency across treatments, sensors in the PS treatment were installed at the same depth. All sensors continuously recorded soil moisture and temperature at 1 min intervals throughout the experimental period.

2.7. Evaluated Variables

2.7.1. Photosynthesis

The head formation stage is the period of maximum water requirement for cabbage [38]. Therefore, the photosynthetic characteristics of cabbage were measured at the head formation stage (on day 30 of the irrigation treatment) using a LI-6400 portable photosynthesis system (Licor. Inc., Lincoln, NE, USA). Five uniformly growing plants from each treatment were randomly selected, and measurements were taken on the 3rd to 5th fully expanded outer leaves (n = 5). Measurements were conducted under a CO₂ concentration (Ca) of 450 μmol·mol⁻¹ and a photosynthetic photon flux density (PPFD) of 1100 μmol·m²·s⁻¹. The measured parameters included the net photosynthetic rate (P_n; µmol·CO₂·m⁻²·s⁻¹), intercellular CO₂ concentration (C_i; μmol·CO₂·mol⁻¹), stomatal conductance (g_s; mol·H₂O·m⁻²·s⁻¹), and transpiration rate (T_r; mmol·H₂O·m⁻²·s⁻¹). In addition, the stomatal limitation index (L_s), instantaneous water use efficiency (WUEi, Equation (6)), and instantaneous carboxylation efficiency (iCE, Equation (7)) were calculated based on P_n, T_r, and C_i according to the method described by Liao [39].

$${\mathrm{L}}_{\mathrm{s}}=1-{\displaystyle \frac{{\mathrm{C}}_{\mathrm{i}}}{{\mathrm{C}}_{\mathrm{a}}}}$$

(5)

$$\mathrm{WUEi}={\displaystyle \frac{{\mathrm{P}}_{\mathrm{n}}}{{\mathrm{T}}_{\mathrm{r}}}}$$

(6)

$$\mathrm{iCE}={\displaystyle \frac{{\mathrm{P}}_{\mathrm{n}}}{{\mathrm{C}}_{\mathrm{i}}}}$$

(7)

2.7.2. Yield and Water Use Efficiency

At harvest, five representative cabbage plants were randomly selected from each treatment (n = 5). The heads were separated and measured for diameter (HD, cm), fresh weight (FW, g), and dry weight (DW, g). Samples were first dried at 105 °C for 24 h and then at 85 °C until reaching a constant weight for DW determination. Both FW and DW were recorded using an electronic scale (MW-2N, CAS Co., Ltd., Yangju, Republic of Korea). Yield was calculated based on head FW and planting density, and expressed as t∙ha⁻¹. The yield water use efficiency (WUE) was calculated according to Equation (8):

$$\mathrm{WUE}(\mathrm{kg}·{{\mathrm{m}}^{-}}^3)={\displaystyle \frac{\mathrm{Y}}{\mathrm{I}\mathrm{D}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{u}\mathrm{a}\mathrm{l}}}$$

(8)

where Y is the cabbage yield (kg·ha⁻¹) and IDactual (m³) is the actual total irrigation depth measured using flow meters.

2.8. Statistical Analysis

Due to the large volume of real-time soil moisture and meteorological data, Python (version 3.8.8) was used to calculate hourly averages. A two-way ANOVA was employed to examine the cabbage growth and production variables. Tukey’s test was used to analyze significant differences (p ≤ 0.05) between treatment means. All statistical analyses were performed using IBM SPSS Statistics (Version 29.0.0, IBM Corp., Armonk, NY, USA), and graphs were plotted using OriginPro 2021 (OriginLab Corporation, Northampton, MA, USA).

3. Results

3.1. Irrigation System Performance

To evaluate the accuracy of on-site meteorological sensors, hourly data from the test site and the Korea Meteorological Administration (KMA) were compared over five days (Figure 4). Key parameters used for reference evapotranspiration (ET₀) calculation showed strong correlations: atmospheric pressure (R² = 0.99), relative humidity (R² = 0.96), solar radiation (R² = 0.99), and temperature (R² = 0.97). Minor systematic deviations were observed, with slightly higher pressure and humidity, and slightly lower radiation and temperature at the test site. Wind speed showed a low correlation (R² = 0.21), likely because of local terrain and microclimatic effects. ET₀ values calculated from sensor data were also consistent with KMA data (R² = 0.97), indicating good overall reliability. These results suggest that, except for wind speed, the sensor data are sufficiently accurate for use in automated irrigation systems.

To evaluate the performance of the automated irrigation system, the calculated irrigation depth (ID) was compared with the actual total irrigation depth (IDactual) recorded by mechanical flow meters for each treatment (Figure 5). The IDactual in both the coir substrate (CS) and paddy soil (PS) treatments were close to the target volumes derived from ETc-based calculations. In the PS treatments, the relative errors ranged from 0.88 to 1.26%, while in the CS treatments, the relative errors ranged from 0.49 to 1.82%. All relative errors were within the stated accuracy of the flow meters (±2.0%), indicating a high level of reliability in the automated irrigation system. These results also indicated that the irrigation system performed consistently and accurately across different growing media and irrigation levels, with only minor variations attributable to allowable measurement errors or slight operational differences.

3.2. Soil Moisture Dynamics

Figure 6 illustrates the hourly solar radiation, cumulative ET₀, soil moisture data, and irrigation events under different irrigation treatments for CS and PS collected from 22 May to 24 May 2024. Solar radiation followed a typical diurnal cycle, reaching zero at night and peaking around noon, indicating that the on-site weather sensor was functioning correctly. The daily cumulative ET₀ values for 22, 23, and 24 May were 5.01, 5.04, and 4.01 mm, respectively. According to the irrigation algorithm, irrigation events were triggered for the PS treatment when the cumulative ET₀ reached 3 mm. Consequently, one irrigation event occurred on 22 May. On 23 May, the cumulative ET₀ from the previous day and continued accumulation caused the 3 mm threshold to be reached twice, leading to two irrigation events. On 24 May, the cumulative ET₀ exceeded the threshold again, prompting one irrigation event. In the CS treatment, an irrigation event was triggered when the cumulative ET₀ reached 1 mm. Based on daily cumulative ET₀ values, five irrigation events occurred on both 22 and 23 May, and four events occurred on 24 May.

Differential soil moisture responses to irrigation events were observed among treatments. In the PS treatment, soil moisture dynamics varied with irrigation treatments. In PS-ETc140, volumetric water content (VWC) ranged from 38.5 to 45.0%. Due to the relatively high irrigation volume per event and the poor drainage of paddy soil, a delayed increase in VWC was observed after irrigation, with notable overall fluctuations. In PS-ETc100, VWC ranged from 36.3 to 38.4%, with a slightly smaller fluctuation range compared with PS-ETc140. In PS-ETc60, VWC ranged from 30.7 to 32.8%, with a further reduction in fluctuation range, indicating that reduced irrigation amount led to more stable moisture conditions. In PS-ETc0, VWC remained within a lower range of 29.8 to 30.8% and showed a slow decline over time. The CS exhibited distinct responses to different irrigation treatments. In CS-ETc140, VWC ranged from 30.9 to 37.6%. High irrigation volumes caused a rapid increase in VWC after irrigation, followed by a gradual decrease, resulting in noticeable fluctuations. In CS-ETc100, VWC ranged from 26.7 to 31.5%, showing slightly less fluctuation compared with CS-ETc140. In CS-ETc60, the VWC range further narrowed to 21.5–25.0%, showing a trend toward more stable moisture conditions with reduced irrigation. In CS-ETc0, VWC remained consistently low, ranging from 2.3 to 2.8%, with minimal fluctuation throughout the period.

3.3. Photosynthetic Performance

The photosynthetic characteristics of cabbage were significantly influenced by growth substrates and irrigation depth (Figure 7). Overall, cabbages grown in CS exhibited higher net photosynthetic rate (P_n), transpiration rate (T_r), and stomatal conductance (g_s) compared with those grown in PS. This indicated that CS was more conducive to photosynthesis. Additionally, CS treatment showed higher instantaneous carboxylation efficiency (iCE), suggesting that CS enhances CO₂ utilization efficiency. The stomatal limitation index (L_s) in CS was generally lower than in PS, indicating that the poor aeration of paddy soil limited stomatal regulation, thereby affecting photosynthetic efficiency. In CS treatment, P_n, T_r, and iCE first increased and then decreased as irrigation depth decreased. The best performance was observed in the CS-ETc100 and CS-ETc60 treatments, which were significantly higher than both the non-irrigation (CS-ETc0) and over-irrigated (CS-ETc140) treatments. In contrast, photosynthetic parameters in the PS treatment were generally lower. However, PS-ETc60 showed significantly higher P_n and iCE compared with the other irrigation treatments. No significant differences in g_s and L_s were observed as a function of irrigation depth in PS, which may be attributed to the limited aeration capacity of paddy soil. Although cabbage grown in CS exhibited enhanced photosynthetic performance, instantaneous water use efficiency (WUEi) did not significantly differ between the two growth substrates and irrigation depths.

3.4. Yield and Growth Traits

The yield and growth traits of cabbage were significantly affected by both growth substrate and irrigation depth (

Table 6. Yield and growth traits of cabbage (Brassica oleracea var. capitata L.) under different irrigation depths (IDs) and grown on growth substrates (GS) (coir substrates-CS and paddy soil-PS).

Treatments	HD (cm)	HFW (g)	HDW (g)	Yield (t·ha−1)	WUE (kg·m−3)
GS
PS	9.8 ± 0.45 b	429.1 ± 54.71 b	27.9 ± 3.64 b	23.9 ± 3.04 b	23.4 ± 3.14 b
CS	13.8 ± 0.44 a	1309.0 ± 103.75 a	84.6 ± 6.76 a	72.8 ± 5.77 a	60.6 ± 8.56 a
ID
ETc140	10.8 ± 0.50 a	603.3 ± 101.41 a	39.0 ± 6.50 a	33.5 ± 5.64 a	17.4 ± 2.91 a
ETc100	13.1 ± 0.84 a	1158.4 ± 203.57a	74.5 ± 12.91 a	64.4 ± 11.32 a	44.2 ± 7.59 a
ETc60	13.1 ± 0.68 a	1080.2 ± 181.00 a	69.9 ± 11.96 a	60.1 ± 10.06 a	64.3 ± 10.98 a
ETc0	10.3 ±1.12 a	634.4 ± 177.23 a	41.4 ± 11.80 a	35.3 ± 9.85 a	-
GS × ID
PS-ETc140	10.0 ± 0.39 c	400.2 ± 60.04 de	26.4 ± 4.55 de	22.3 ± 3.34 de	11.6 ± 1.74 c
CS-ETc140	11.7 ± 0.75 bc	806.3 ± 148.53 cd	51.6 ± 9.51 cd	44.8 ± 8.26 cd	23.2 ± 4.27 bc
PS-ETc100	11.0 ± 0.47 bc	606.4 ± 73.85 de	38.7 ± 4.75 de	33.7 ± 4.11 de	23.7 ± 2.89 bc
CS-ETc100	15.2 ± 0.90 a	1710.3 ± 169.38 a	110.3 ± 9.30 a	95.1 ± 9.42 a	64.7 ± 6.41 ab
PS-ETc60	11.3 ± 0.48 bc	595.5 ± 86.21 de	39.0 ± 6.22 de	33.1 ± 4.79 de	34.7 ± 5.03 bc
CS-ETc60	14.8 ± 0.52 a	1564.8 ± 150.13 ab	100.8 ± 11.29 ab	87.0 ± 8.35 ab	93.8 ± 9.00 a
PS-ETc0	7.1 ± 0.63 d	114.3 ± 24.85 e	7.2 ± 1.64 e	6.4 ± 1.38 e	-
CS-ETc0	13.5 ± 0.39 ab	1154.5 ± 73.93 bc	75.6 ± 6.26 bc	64.2 ± 4.11 bc	-
GS	***	***	***	***	***
ID	***	***	***	***	***
GS × ID	**	*	*	*	***

HD—head diameter; HFW—head fresh weight; HDW—head dry weight; WUE—water use efficiency. The treatments consisted of four irrigation depths based on: ETc140 (140% of ETc), ETc100 (100%), ETc60 (60%), and ETc0 (no irrigation control). All data are expressed as mean ± standard error (n = 5). Different letters indicated significant differences (p < 0.05) according to Tukey’s multiple range test and t-test. NS, *, **, and *** represent non-significant or significant at p < 0.05, 0.01, and 0.001, respectively.

). Overall, cabbage grown in CS exhibited significantly higher head diameter, fresh weight, dry weight, total yield, and water use efficiency (WUE) compared with PS. When comparing irrigation depth independently, no significant differences were observed among them. Within the PS treatments, the effect of irrigation depth on yield and growth traits was minimal, with PS-ETc100 showing a slight advantage over the other treatments. In contrast, within the CS treatments, both CS-ETc100 and CS-ETc60 increased head diameter, fresh and dry weight, compared with CS-ETc140 and CS-ETc0. Correspondingly, total yield followed a similar trend, with CS-ETc100 achieving the highest yield and CS-ETc60 slightly less, although no significant difference was observed between them. WUE was also significantly influenced by both growth substrates and irrigation depth. The highest WUE was observed in CS-ETc60, followed by CS-ETc100, suggesting that moderate deficit irrigation under CS conditions was beneficial for optimizing water use. In contrast, WUE in PS remained relatively low and showed no significant differences among irrigation depths. Additionally, the interaction between growth substrates and irrigation depths significantly affected all yield and growth traits, indicating that cabbage responses to irrigation strategies differed by root-zone conditions. These results suggest that irrigation optimization strategies should be adapted to the specific characteristics of the growth substrates. Morphological observations in Figure 8 showed that cabbage heads grown in CS-ETc100 and CS-ETc60 were not only larger but also firmer. Therefore, future crop management should optimize irrigation by considering both the dynamic water requirements of plants and the properties of the growth substrate, aiming to maximize yield and water use efficiency.

4. Discussion

4.1. Functionality and Effectiveness of the Automated Irrigation System

In this study, we developed a precise irrigation system based on real-time weather sensing and the cumulative reference evapotranspiration (ET₀) concept. Among the parameters used for calculating ET₀, the on-site wind speed data showed some discrepancies compared with KMA data (Figure 4). In our previous research, we also applied the XGBoost model to predict weather parameters for ET₀ calculation, but the wind speed prediction showed a relatively low R² value (0.730) [40]. Previous studies have demonstrated that wind speed exhibits strong spatial variability, with significant differences possible even over short distances [41,42]. This variability is influenced by multiple factors such as surrounding buildings, topography, and microclimate, resulting in discrepancies in wind speed data among different measurement sites [43]. Moreover, the contribution of wind speed to ET₀ calculation exhibits significant seasonal variation. During summer, ET₀ shows the lowest sensitivity to wind speed, while solar radiation and temperature exert greater influence [44,45]. Conversely, ET₀ responds more strongly to wind speed in winter [46]. Therefore, the installation of on-site weather sensors remains essential for precise irrigation management based on the Penman–Monteith equation.

Irrigated agriculture uses about 70% of the world’s available freshwater each year, yet irrigates only around 25% of cultivated land [47]. Consequently, the concept of ‘more yield per water drop’ (MYWD) has emerged as a priority for ensuring water security and sustainable food production. However, the adoption of advanced water-saving technologies remains limited in low-income countries [48]. In some regions, irrigation scheduling is still based on manual measurements and empirical judgment, which can result in improperly timed irrigation that fails to match the crop’s critical water demand periods, thereby leading to water wastage and yield loss [49]. Currently, increasing research efforts are directed toward developing low-cost, efficient precision irrigation technologies [13]. Most automated irrigation systems rely on real-time soil moisture monitoring for feedback control. However, their operational stability is often limited by the spatial distribution and accuracy of the sensors, particularly in drip irrigation systems, where non-uniform wetting can lead to inaccurate soil moisture readings [31,50]. Remote sensing technologies have shown potential in large-scale irrigation monitoring, particularly in irrigation detection and quantitative estimation. Nevertheless, their application remains limited by low spatial resolutions, long revisit intervals, and significant data uncertainties, making it difficult to achieve precise irrigation management at the field scale [51]. In contrast, the system developed in this study enables continuous and dynamic estimation of crop water requirements and allows precise control of irrigation timing and volume at the plot scale by adjusting the Kc value in real time according to the target crop and its growth stage.

4.2. Effects of Automated Irrigation System on Soil Moisture Dynamics

Considering the significant differences in physical properties between PS and CS, different cumulative ET₀ thresholds were adopted in this study to adjust irrigation frequency accordingly. CS exhibited better aeration and a more pronounced and rapid increase in soil moisture after irrigation, whereas PS showed a more muted and stable soil moisture response (Figure 6). These differences in moisture dynamics reflect the fundamental disparity in drainage performance between the two substrates. Previous studies have demonstrated that the plow sole layer in PS has the lowest permeability, with a hydraulic conductivity significantly lower than that of the surface and subsoil layers, thereby restricting vertical water infiltration [52,53]. As a result, high-frequency and low-volume irrigation in PS maintains a persistently saturated surface layer, inhibiting oxygen diffusion and creating hypoxic or even anaerobic conditions in the root-zone [54,55]. Such conditions are particularly detrimental to shallow-rooted vegetables such as cabbage, often resulting in leaf chlorosis and root rot [56]. Additionally, the high clay content in PS leads to rapid surface evaporation and crusting under high temperatures, further limiting plant growth [57,58]. In contrast, CS has a higher water-holding capacity and moderate capillarity, enabling it to maintain a stable root-zone water potential without causing waterlogging [59]. High-frequency and low-volume irrigation helps to reduce soil moisture fluctuations and enhance water use efficiency [60,61]. Previous research has applied irrigation scheduling based on cumulative crop evapotranspiration (ETc), combined with soil sensor-controlled irrigation [60], which shows methodological similarity to this study. The key distinction of this study is the use of cumulative ET₀ as the irrigation trigger, enabling dynamic adjustment of Kc values and irrigation thresholds based on crop growth stages. Our irrigation system provides enhanced flexibility and broader applicability, supporting the management of diverse crops and practical implementation.

4.3. Cabbage Photosynthesis

The interaction between growth substrates and irrigation depth influences cabbage photosynthetic performance, likely because of differences in root-zone aeration (Figure 7). Enhanced root-zone aeration has been reported to improve the effective quantum yield of photosystem II (ΦPSII), facilitating photosynthetic electron transport and increasing CO₂ assimilation through the Calvin–Benson cycle [62]. In this context, CS likely provides a more favorable root environment than PS, supporting higher photosynthetic efficiency. These findings are consistent with a previous report showing that root-zone oxygen supply can enhance photosynthetic performance in tomato [63]. Although CS treatments showed significantly higher g_s, the L_s did not differ significantly between the two substrates under ETc140, ETc100, and ETc0 irrigation treatments. This suggests that factors beyond stomatal regulation, such as alleviation of non-stomatal limitations, contributed to improved carbon assimilation. Research indicates that better root-zone aeration supports photosynthesis by improving stomatal behavior and alleviating non-stomatal limitations such as Rubisco activity, thereby enhancing CO₂ assimilation [64,65,66]. Notably, although both P_n and T_r were significantly higher in CS than in PS under the same irrigation depth, the instantaneous water use efficiency (WUEi) showed no significant variation. This response may reflect a leaf thermal regulation strategy [67]. Under high light and moderate water conditions, plants may enhance transpiration to cool leaf surfaces, thereby preserving the activity of heat-sensitive enzymes such as Rubisco [68]. Different irrigation depths also significantly influenced the photosynthetic performance of cabbage. The P_n and T_r reduction in ETc140 is likely due to large soil moisture fluctuations disrupting root-zone oxygen diffusion and soil redox potential (Eh), which impair root respiration and nutrient availability [69,70]. Such conditions can limit key biochemical processes involved in carbon fixation, reducing photosynthetic capacity [66,71]. These results highlight the importance of precise irrigation management to minimize moisture fluctuations and sustain crop productivity across substrates.

4.4. Interaction Effects on Cabbage Yield

The efficiency of photosynthetic carbon assimilation plays a crucial role in determining biomass accumulation and crop yield [72]. In this study, the trends in photosynthetic performance were closely reflected in yield performance. Both GS and IL had highly significant effects on cabbage yield (p < 0.001), and their interaction was also significant (p < 0.05). Improved aeration and water availability in CS treatments likely enhanced CO₂ assimilation efficiency, facilitating head formation and contributing to increased fresh yield, as supported by previous studies [73]. Conversely, poor drainage in PS induced root-zone hypoxia, which has been shown to delay vegetative development and reduce head growth [74,75]. These physiological constraints may explain the significantly lower yields observed under PS treatments. Despite higher irrigation in CS-ETc140, yield decreased, likely because of excessive moisture causing root hypoxia, which limits photosynthesis and reduces biomass and head growth [11,75]. In both substrates, a yield reduction was observed under the ETc0 treatment. This was mainly caused by stomatal closure and reduced photosynthesis resulting from drought stress [76]. Previous studies have shown that moderate irrigation promotes the allocation of assimilates and improves head firmness [77]. These results underscore the importance of balancing substrate properties and irrigation levels to optimize cabbage growth and maximize yield.

5. Conclusions

This study demonstrated that reusing discarded coir substrates in paddy fields, in combination with an automated ET₀-based irrigation system, can substantially improve water use efficiency, optimize root-zone moisture conditions, and enhance cabbage yield. The proposed high-frequency, low-volume irrigation approach maintained stable performance across different irrigation levels and substrates, providing a practical alternative to conventional drainage systems. The results showed that the CS-ETc100 treatment achieved the highest yield, whereas the CS-ETc60 treatment provided the maximum WUE with no significant yield reduction. By integrating organic substrate reutilization with precision irrigation, this strategy offers a sustainable and adaptable solution for paddy field conversion and holds strong potential for wider adoption in multi-crop production systems.