Introduction
Korea’s rural population is aging rapidly, and the resulting labor shortage in the agricultural sector is becoming increasingly severe. Notably, approximately 78% of farmers cultivating flowers and ornamental crops, including chrysanthemums, are over 60 years old and continue to rely on labor-intensive cultivation practices, leading to low production efficiency (MAFRA, 2024). Chrysanthemum (Chrysanthemum x morifolium) is a crop that requires precise water management to achieve optimal growth and quality (Morrison, 2024). Despite the advantages of hydroponic cultivation in enabling more precise control, high initial costs associated with facility construction hinder its widespread adoption. Consequently, approximately 90–95% of cut flower chrysanthemum growers still utilize conventional soil-based methods (Oh et al., 2024).
In recent years, sensor-based automated irrigation systems have been increasingly adopted across various crops and are considered a core technology for advancing agricultural automation. Among these, systems incorporating soil moisture sensors offer the advantage of real-time irrigation control based on crop water requirements, thereby reducing labor inputs and improving resource efficiency (Montesano et al., 2018). However, most existing studies have focused primarily on soilless or substrate-based cultivation systems (Vera et al., 2021), and research on sensor-based irrigation systems optimized for soil-based cultivation remains limited.
Accordingly, this study aims to evaluate the suitability of soil moisture sensors for accurate water status monitoring under conventional soil cultivation conditions.
Materials and Methods
Plant and growth conditions
This study was conducted in two standard-type chrysanthemum greenhouses located in Gangseo-gu, Busan, Korea. Although both greenhouses were equipped with fertigation systems (WIN-5000S, WOOSUNG HIGHTEC CO., LTD., Gyeongsangnam-do, Korea), irrigation and fertigation were manually applied based on each farmer’s decision, taking into account the specific soil conditions on each farm. To minimize environmental variability, two greenhouses approximately 200 meters apart were selected: Farm A (35.2111°N, 128.9268°E) and Farm B (35.2120°N, 128.9290°E) (Fig. 1A). The chrysanthemum cultivar used was ‘Baekgang’, bred in Korea (Fig. 1B). Uniform plants were transplanted around September 20, 2024, and cultivated for approximately 150 days until the harvest of the first flower.
Atmosphere and soil environmental monitoring
To monitor both aboveground and belowground environments in real time, a data logger (ZL6, METER Group, Pullman, WA, USA) was employed. An air temperature and humidity sensor (ATMOS 14, METER Group) and a quantum sensor (SQ-521, Apogee Instruments, Logan, UT, USA) were connected to the logger to measure air temperature, relative humidity, vapor pressure deficit (VPD), and photosynthetic photon flux density (PPFD).
For soil condition monitoring, the same logger was connected to soil moisture and electrical conductivity (EC) sensors (TEROS 12, METER Group) using the FDR method to measure volumetric water content (VWC; %) and bulk EC (mS·cm-1). For accurate VWC measurement, a calibration equation (θ = 3.88 × 10-4 × sensor output − 0.696) was applied, considering soil hydraulic properties. Bulk EC values were converted to pore water EC using the Hilhorst equation (Hilhorst, 2000). Additionally, a soil matric potential (SMP) sensor (TEROS 21, METER Group) was used to assess matric potential (kPa). All sensors were installed at a 5 cm depth to target the effective rhizosphere, where water and nutrient uptake are most active, following the manufacturer’s guidelines. Two replicates of each sensor type were installed in each greenhouse.
Flowering quality and data analysis
To evaluate flowering quality, at least three flowers were harvested from each farm. Growth characteristics such as plant height (from the base to the apical tip) and peduncle length (from the bract to receptacle), peduncle thickness were measured at harvest. The flowers were subsequently placed in a preservative solution containing FloraLife® 200 (Smithers-Oasis Company, SC, USA) and maintained at room temperature (24.0 ± 1.1°C) and relative humidity (60.5±6.2%) until full bloom. At that stage, flowering quality characteristics such as capitulum (inflorescence) diameter and height were measured.
An independent two-sample t-test was conducted to compare flowering quality between two greenhouses. Prior to each t-test, homogeneity of variance was assessed using the folded F-test. When the assumption of equal variances was met (Pr > F > 0.05), pooled variance estimates were used; otherwise, the Satterthwaite approximation was applied. All analyses were performed using SAS software (version 9.4; SAS Institute Inc., Cary, NC, USA). Graphs were generated using SigmaPlot (ver. 11.0; Systat Software Inc., USA).
Results and Discussion
During the cultivation of Chrysanthemum ‘Baekgang’ in the two greenhouses, average air temperature was 18.3 ± 2.9°C in Farm A and 16.8 ± 2.6°C in Farm B. Relative humidity averaged 85.3 ± 13.6% and 85.4 ± 10.6%, respectively, indicating similar atmospheric conditions (Fig. 2). According to RDA (2024), the temperature and RH in both farms were within the optimal range for chrysanthemum production. VPD, a key parameter affecting plant transpiration, was 0.37 ± 0.3 kPa in Farm A and 0.32 ± 0.3 kPa in Farm B. The average PPFD was 359.6 μmol·m-2·s-1 in Farm A and 355.2 μmol·m-2·s-1 in Farm B. These results confirm that ambient conditions were similar across the two sites during the winter season (Fig. 3).
VWC averaged 23.0 ± 0.2% in Farm A and 26.2 ± 0.3% in Farm B, with a consistent difference of about 3% (Fig. 4). Maximum VWC values were 26.9% in Farm A and 29.6% in Farm B, while minimum values were 20.8% and 22.2%, respectively. In contrast, SMP showed clearer differences, with an average of −10.2 ± 2.5 kPa in Farm A and −6.8 ± 1.6 kPa in Farm B. Based on the actual changes in SMP observed during the experiment, Farm A maintained SMP near the water saturation level (0 kPa) for approximately 60 days after planting, and subsequently kept the SMP within the range of field capacity (-10 to -33 kPa) until harvest. In contrast, Farm B maintained water saturation conditions for about 100 days after treatment, after which the SMP gradually declined to approximately -51 kPa until harvest. The differing irrigation strategies between the two farms, as monitored through SMP measurements, were attributed to variations in soil electrical conductivity (EC). Farm B exhibited a very high pore-water EC of 8.25 mS·cm-1 due to salt accumulation in the soil where chrysanthemums were cultivated (Fig. 5). Consequently, saturated irrigation was applied for approximately 100 days to mitigate growth inhibition caused by salinity stress, which resulted in a reduction of pore-water EC to about 3.5 mS·cm-1 by day 100.
In recent years, the development of automated irrigation systems based on soil moisture sensors have been developed to improve the water use efficiency and crop quality in greenhouse cultivation for horticultural crops. Among these, systems utilizing FDR or capacitance sensors have gained attention due to their cost-effectiveness. While these sensors are primarily applied in soilless cultivation using containers (Wheeler et al., 2018), efforts are also underway to develop irrigation systems employing FDR sensors for soil-based cultivation in open field and greenhouse (Abdelmoneim et al., 2025;Martínez-Gimeno et al., 2020). Kang et al. (2021) reported that when SMPs ranged from –10 kPa to –100 kPa, the corresponding VWC measured using FDR sensors varied by as little as 5% and up to 12%. These findings are consistent with the results of the present study, which demonstrated that changes in VWC measured by the FDR sensor were smaller than the corresponding changes in SMP under the same moisture conditions in soil. The observed variation falls within the typical ±3% measurement error of FDR sensors. Since the range of change is comparable to the sensor’s accuracy threshold, the validity of the measurements is uncertain. Therefore, the applicability of FDR sensors under these experimental conditions should be carefully reconsidered. This is consistent with findings by Kim et al. (2021), who reported that matric potential sensors provide approximately 6.6 times higher resolution than FDR sensors at comparable moisture levels.
When evaluating the flowering quality of chrysanthemums, the thickness of the peduncle, an important factor for postharvest quality, was significantly greater on Farm A (6.9 ± 0.1 mm) than on Farm B (5.7 ± 0.2 mm), with the difference being statistically significant (p = 0.0143) (Table 1). Similarly, the capitulum diameter, another indicator of flower size, was significantly larger (p = 0.0307) in chrysanthemums grown on Farm A (106.6 ± 6.9 mm) compared to those from Farm B (77.4 ± 4.7 mm). Although the capitulum height, another measure of flower size, did not show a statistically significant difference, chrysanthemums from Farm A exhibited a 40% greater height (38.5 ± 4.7 mm) than those from Farm B (27.5 ± 2.2 mm). In addition, the number of days to flowering was significantly shorter in Farm A (109 days) than in Farm B (150 days).
Throughout the cultivation period, above-ground environmental conditions were similar between the two farms. However, differences in soil moisture and management led to significant differences in flowering quality. In particular, the reduced flowering quality observed in Farm B is likely attributable to elevated soil salinity and prolonged over-saturated conditions.
To develop an effective automated irrigation system for soil-based chrysanthemum cultivation, SMP are more appropriate than VWC sensors due to their greater sensitivity to soil water status. Furthermore, integrating soil EC sensors with chemical property analyses is essential for managing nutrient and salinity levels. Such an approach would enable more precise irrigation control and enhance crop quality in cut chrysanthemum production.
