**4. Materials**

The functionality and reliability of the whole system were proven during two continued experiments with two different natural soils, characterized by very different soil hydraulic properties (see Table 2).

**Table 2.** Main hydraulic properties of study soils. *Ks* = saturated hydraulic conductivity; *θs* and *θr* = saturated and residual water content, respectively; *bd*= bulk density.


The fine-textured soil (a Silty Loam, according to the United States Department of Agriculture, USDA, classification [68,69]) was composed of 1% gravel, 22% sand, 54% silt, and 23% clay (Figure 10a), while the coarse-textured soil (a Loamy Sand, according to USDA) was composed of 4% gravel, 79% sand, 11% silt, and 6% clay (Figure 10b).

**Figure 10.** Grain size distribution of the soils used for the experiments: (**a**) fine-textured soil (Silty Loam, according to USDA); (**b**) coarse-textured soil (Loamy Sand, according to USDA).

#### **5. Methods, Tests, and Results**

Focusing on the Plant Nodes, the system has been tested during a continued experiment where the two different greenhouse soils were watered several times, to verify if the sensor was able to reliably acquire, transmit, and store the ambient temperature and the soil water content parameters in real time and to show them on the custom GUI.

The measurements were made in a plastic box initially filled with expanded clay aggregate which allowed percolated water to outflow and where a Sentek Drill & Drop Probe (hereafter named "reference sensor") was driven (Figure 11a). Then the remaining top 30 cm of the box was filled with the chosen soil, either Loamy Sand or Silty Loam. Both soils were packed in 0.05 m lifts and gently tapped into place. This accurate packing mechanism was adopted to achieve homogeneity vertically, to keep perfect contact at the interface, and to minimize preferential flow along the sides of the box. The reference sensor was placed at the center of the box. It features an array of water content and temperature sensors placed at 5 cm, 15 cm, 25 cm, 35 cm, and 45 cm from the top surface.

**Figure 11.** The plastic box used for our experiments. (**a**) Expanded clay aggregate bottom filler, together with a Sentek Drill & Drop Probe. (**b**) Plant Node #1 and Sentek sensor during acquisition.

Then our Plant Node #1 was inserted in the soil at a distance of 10 cm from the Sentek sensor (Figure 11b, where Node #1 is shown without the lid and connected to a 230 ACV-5 DCV adapter during a test measurement). Since the reference sensor and the Plant Node #1 are installed in different positions/depths, this has an impact on the measurements, as explained in the following sections. Data of Plant Node #1 were collected every 5 min for several days while the automatic acquisition system of the reference sensor stored the measurement results every minute. In carrying out the measurements, the soil was watered in consecutive steps.

Before and after each measurement a calibration was performed on the Capacitive Soil Moisture Sensor v1.2 measuring water content, exposing it for 15 min. to air, then dipping it for 15 min in tap water. The reproducibility of these measurements certifies that the low-cost water content is in working order.

#### *5.1. Measurements in Silty Loam*

In Figure 12 we compare the water content measured by the reference sensor at a depth of 5 and 15 cm in Silty Loam. After installing the sensors in a uniformly and slightly moistured Silty Loam (initial volumetric water content of 10%), then four synchronous waterings, clearly visible at a depth of 5 cm, were performed during the last two days of this measurement. The two plots witness the strong dependence of the water content on the soil depth in Silty Loam.

**Figure 12.** Water content measured by the reference system at two different depths in Silty Loam: red, 5 cm underground, and green, 15 cm underground.

Figure 13 shows the water content measured by the reference system (5 cm underground) and the output voltage of the Capacitive Soil Moisture Sensor v1.2 in Silty Loam (Node #1). The qualitative correlation between the two plots is evident: each watering causes an increase in the measured water content of the reference sensor and a decrease in the output voltage of Node #1. Moreover, the long time elapsed in soil with VWC of about 10% before the first watering guarantees the Capacitive Soil Moisture Sensor v1.2 had plenty of time to reach its settling time. However, we note the lack of linearity of the Capacitive Soil Moisture Sensor v1.2: its sensitivity is too high for small values of water content and it is substantially reduced for volumetric water contents greater than about 15%. The low draining capability of this soil which maintains its water content during the time causes high values of water content and for this reason, the Capacitive Soil Moisture Sensor v1.2 works most of the time almost in saturation. Future improvements for the sensor should be directed towards the linearization of the input/output curve to obtain a constant sensitivity. A detailed discussion of the correlation between the results of the two sensors is reported below.

**Figure 13.** Water content measured by the reference system (red, 5 cm underground) and the Capacitive Soil Moisture Sensor v1.2 (blue) in Silty Loam. Initial and final peaks of the Node #1 plot represent a calibration of the Capacitive Soil Moisture Sensor v1.2 obtained by placing the sensor for 15 min in the air (maximum peak) and 15 min in tap water (minimum peak).

Figure 14 shows the reference temperature compared to the temperature of Node #1. In addition, in this case a slight difference is detected, most likely due to the distance of the two sensors and the intrinsic measurement error. Indeed, the LM35 declares a 0.5 ◦C ensured accuracy (at 25 ◦C) while the reference Sentek system has a temperature error of 0.1 ◦C.

#### *5.2. Measurements in Loamy Sand*

In Figure 15 we compare the water content measured by the reference sensor at a depth of 5 and 15 cm in Loamy Sand. The water content curve at 5 cm clearly shows five consecutive waterings performed during the 2 days of this measurement. On the other hand, the water content curve at 15 cm shows an increase only after the 3rd watering, clearly witnessing the dependence of water content on the soil depth. Furthermore, due to the alternation of rainfall and water redistribution periods, the evolution in time of the wetting front is very complex, as [70,71] showed in their schemes with compound profiles.

Figure 16 shows the water content measured by the reference system (5 cm underground) and the output voltage of the Capacitive Soil Moisture Sensor v1.2 in Loamy Sand. In addition, for this soil, we obtain a "first sight" reasonable qualitative agreemen<sup>t</sup> between the results of the two sensors. Again, we note that the sensitivity of the Capacitive Soil Moisture Sensor v1.2 is too high for small values of water content and it is substantially reduced

for volumetric water contents greater than about 10% for this soil material. A detailed discussion of the correlation between the results of the two sensors is reported below.

**Figure 14.** Temperature measured by the reference system (red, 5 cm underground) and the LM35 mounted in Node #1 (blue) in Silty Loam. The rectangular area shows the region where we realized the calibrations of Section 6.2.

**Figure 15.** Water content measured by the reference system at two different depths in Loamy Sand: orange, 5 cm underground, and blue, 15 cm underground.

**Figure 16.** Water content measured by the reference system (orange, 5 cm underground) and the Capacitive Soil Moisture Sensor v1.2 (blue) in Loamy Sand. Initial and final peaks of the Node #1 plot represent a calibration of the Capacitive Soil Moisture Sensor v1.2 obtained by placing the sensor for 15 min in the air (maximum peak) and 15 min in water (minimum peak).

Figure 17 shows the reference temperature compared to the temperature of Node #1. A slight difference is detected, most likely due to the distance of the two sensors and the intrinsic measurement error.

**Figure 17.** Temperature measured by the reference system (orange, 5 cm underground) and the LM35 mounted in Node #1 (blue) in Loamy Sand.
