**2. Proximal soil sensing**

Facing the growing demand for food and sustaining water resources needs irrigation optimization employing advanced technologies such as soil moisture sensors [53]. Technologies such as drip watering, proximal sensors, and remote controllers for water management have joined the farming sector owing to agricultural development and subsequently rising demand for freshwater [54]. Considering that implementing a systematic irrigation plan for farmers is practically complicated, digital instruments effectively assist in accurate irrigation planning [55]. Furthermore, the proximal platform can be used to evaluate plant health [56]. Recent advances in electromagnetic moisture sensor technologies have facilitated automatic irrigation

#### **Figure 2.**

*Positive and negative attributes of proximal and aerial sensors.*

### *DOI: http://dx.doi.org/10.5772/intechopen.103898 An Overview of Soil Moisture and Salinity Sensors for Digital Agriculture Applications*

scheduling [57], which enhances water-use efficiency. These sensors are divided into active and passive instruments, which are applied for crop yield assessment and watershed management in digital agriculture [58]. In another classification system, soil sensors can be divided into resistive or capacitive sensors. Resistance-based sensors are easy to use and inexpensive. However, error sources affect their accuracy and efficiency [59]. Jusoh et al. [60] reported that the resistive sensor operates defectively in sandy loam and clay loam soils owing to low bulk density and high organic matter.

As efficient machines, capacitive soil moisture sensors are affordable for reducing water costs and wastage and computerized scheduling of irrigation [57, 61]. Capacitive probes and electronic TDR soil moisture sensors with *in situ* measurement have easy use, high accuracy, and fast data retrieval that are extensively used to monitor soil moisture changes in fields and predict drought, particularly in arid and semiarid lands. Furthermore, these instruments are applied for hydrological flux calculations, modeling runoff infiltration, and calibration of remote sensing data. However, precisely estimating moisture content is not convenient due to the spatial diversity of soils and the spatiotemporal heterogeneity of soil water content at high depths [36]. It is further challenging to measure moisture using discrete and wirebased instruments in fields with high vegetation diversity and different hydrological properties, which cause numerous obstacles in analysis and control systems, specifically at broad geographical scales [55, 62]. Since some sensors retrieve various data from a farm, it is not possible to automatically turn on or off the federal irrigation system. Moreover, many users have reported fractures of the watermark rod during dipping or separating it from the soil (**Figure 3**). The low accuracy of some sensors, which have a high moisture detection limit and detect the soil as dry, directly challenged farmers. Therefore, there is a possibility of flooding the root zone and loss of plants in the event of inadequate knowledge of farmers. Hence, growers' propensity to purchase sensors decreases. The cost of sensors determines their resistance and precision in heterogeneous ambient conditions [63]. A flawless calibration process is necessary to optimize the sensor's accuracy. In order to improve the accuracy of the soil moisture sensor, Gonzalez-Teruel et al. [64] calibrated it on three different types of soil. According to Radi et al. [65], the soil moisture sensor SKU:SEN0193 is a low-cost commercial sensor that must be calibrated before being used on farms. **Figure 4** shows the calibration process of a soil moisture sensor. Since different raw materials are used to make sensors, low-cost sensors have low resistance to

**Figure 3.**

*Instances of different soil moisture sensor probes that are used for digital farming applications.*

**Figure 4.** *Calibration of soil moisture sensor for different types of soil. Source: SunBot.de.*

adverse environmental conditions such as sunlight, strong winds, and wild animals. Therefore, it is challenging to achieve integrated systems on farms owing to the natural obstacles. A proximal network is high-priced due to the need for periodic servicing of sensor portions [66], which increases costs for producers. Given that experimental determination of soil moisture is a fundamental characteristic of agricultural operations [66, 67], cost-effective analysis of soil volumetric water content (VWC) is an important strategy for promoting sustainable agriculture through the use of computerized machines and Internet of things (IoT) development, particularly for smallholder farmers [68].

Significant advances have been made in technologies for assessing, mapping, and spatiotemporal monitoring of salinity on a field, regional, and national scale [10]. Generally, there are five methods for estimating salinity on a farm: (1) observing salts on the soil surface, (2) estimating EC in saturated soil extracts, (3) measuring *in situ* electrical resistance, (4) determining *in situ* EC by TDR, and (5) noninvasive EC measurement using EM sensors [69]. The EM38 sensor is one of the most popular sensors in agriculture and consists of a receiver and a transmitter coil with a distance of 1 meter from each other, which are connected at the opposite end of a nonconducting rod, which measures salinity and other soil properties such as nutrient level and clay bulk [70]. This sensor is comfortable to use, and users can interpret its data after processing obtained images [71]. Slavich et al. [72] and Guo et al. [73] used EM38 data to determine soil salinity and barley tolerance to salinity and for digital soil mapping of spatiotemporal salinity changes. Hammam and Mohamed [74] mapped the spatial pattern of soil salinity in the East Nile Delta using geographic information system (GIS) and inverse distance weighting (IDW) techniques. Ding and Yu [75] reported that the obtained EC data from the EM38 sensor were significantly correlated with the experimental soil analysis in the laboratory. Guo et al. [76] recognized a significant correspondence between actual soil EC and sensor data (r > 0.9) by employing EM38 proximal technology. Additionally, EM38 is beneficial for prompt soil assessment before planting operations (**Figure 5**) [77]. Despite the speedy operation of this sensor, its vulnerability to metals and electromagnetic noise sources, such as power cables, can generate fluctuations in data registration [78]. The framework of the soil moisture proximal measuring procedure is outlined in **Figure 6**, along with three models of soil sampler robots.

The integrated wireless sensor network (WSN) is designed to measure soil salinity and support automated irrigation systems [83]. With the WSN, numerous facilities are provided such as remote monitoring of soil fertility, crop water situation, and assistance to the irrigation system with reasonable costs, low energy consumption,

*DOI: http://dx.doi.org/10.5772/intechopen.103898 An Overview of Soil Moisture and Salinity Sensors for Digital Agriculture Applications*

#### **Figure 5.**

*Instances of different portable pH and EC meters used for measuring soil salinity.*

#### **Figure 6.**

*Summary of the soil moisture measurement process by the proximal sensor and models of soil sampler robots. Adapted from [79–82].*

and extended life [84, 85]. In a study by Sui and Baggard [86], WSN sensors automatically recorded soil condition data over the Internet every minute. The combination of WSN with the GIS in a study by Zhang et al. [87] proposed a soil moisture distribution map for accurate irrigation control. This system improves irrigation efficiency by decreasing freshwater loss and watering costs [88]. The precision of the data retrieved by the WSN depends on the system's capability to hold the input voltage constant and the dependability of the calibration curves [89]. Though the WSN with fast data retrieval capability is a promising strategy in precision agriculture, barriers such as soil and canopy interference can affect data validity [79].
