**2. Precision agriculture and its practices in cotton production**

Cotton agriculture is known for its intensive crop management practices with higher levels of input including seeds, land management practices, agrochemicals (fertilizers, pesticides, and herbicides), and water, which is negatively influencing

## *A Review of the Factors Affecting Adoption of Precision Agriculture Applications in Cotton… DOI: http://dx.doi.org/10.5772/intechopen.114113*

farmers' profit and yield [14]. Also, cotton crop is host to many insects and pests, and enormous amounts of pesticides are used which are hazardous for the environment in the long run, in order to control these insects and pests [15]. When PAT is applied to these management practices, it will be possible to improve the economic and environmental sustainability of cotton production.

Several factors, such as the differences in soil texture and fertility and the occurrence of pests, diseases, and nematodes, can cause spatial variability in the growth and development of cotton crops within a field. PA, which comes from the spatial variability distinctive to each field can be measured and managed with site-specific techniques according to the needs of the crop, PA contributes to more effective use of inputs such as fertilizers, pesticides, tillage, and irrigation water. Effective use of inputs will increase crop yield and (or) quality without polluting the environment [7].

To understand fully how PAT is applied, the tools and techniques that create the infrastructure of this modern form of agricultural management need to be well explained. A diagrammatic representation of the PA components is shown in **Figure 1**. It can be said that PA is mainly composed of agricultural data collecting, data processing and analysis, data interpretation and decision making, and variable rate application of inputs [8, 16–18].

The data collecting, which is the first step of PA, entails gathering as much data as possible about crops, soil, fields, terrain, climate, variables, and resource availability by sensors and innovative techniques. Data collecting can be performed using either proximal sensing or remote sensing techniques by special equipment and software such as sensors, GPS technology, controllers, gateways, drones, satellites, and imaging [8].

Data preprocessing and analysis allow for making accurate decisions during variable rate application of inputs by farm machinery because it contributes to a better understanding of crop dynamics, weather, and soil conditions [19]. Information for data interpretation, decision making, and implementation of crop management practices at an appropriate scale and time can be achieved by preprocessing and transforming the raw data acquired through sensing techniques and GIS software [20–22].

#### **Figure 1.** *The main components of precision farming [8, 16, 17].*

Data interpretation and decision-making covers choosing the appropriate management tools which give good outcomes in available natural conditions like soil, and environment before using variable-rate devices installed on agricultural equipment. The processes in this phase are considered as very important step in PA [23]. Tantalaki et al. [24] reported that the high volume and complexity of the data caused challenges in successfully implementing PA. They also emphasized that analyzing and interpreting data obtained from ground sensors, unmanned systems, or remote sensing satellites is a significant issue in the successful implementation of PA. The authors state that machine learning techniques, artificial neural networks, support vector machines, decision trees, and random forests, frequently applied for agricultural management purposes, seem promising to cope with agricultural big data, but need to reinvent themselves to meet existing challenges. Also, image segmentation technologies play a significant role in data interpretation and decision making such as plant or weed identification, crop growth stage prediction, crop disease, row detection, and cotton detection in the field [25]. Zhang et al. [26] developed a software used on the smartphone for the real-time detection of cotton diseases and pests in the field. They stated that the developed software could effectively detect the infected area of cotton leaves in the field and provide a technical support for controlling cotton diseases and pests.

Managing field operations can be performed using the information acquired as decision support. Many crop management practices (multiple fertilizer applications, multiple plant growth regulator applications, multiple irrigation potential applications, multiple pesticide applications, etc.) are intensively applied while growing cotton. For this purpose, several agricultural machinery and tools are used for seedbed preparation, sowing, fertilization, pest control weeding, irrigation, and harvesting to reduce labor costs and increase productivity. In PA, these agricultural machinery and tools manage field operations by using technologies such as IoT, AI, remote sensing, and ImP. Recently, numerous robotic systems and variable-rate applicators with human-like capabilities (e.g., precision spraying systems, harvesting robots, shearing robots, grafting machines, weed control systems, transplanting machines, and path planning) have been developed to manage different agricultural activities such as planting, inter-row cultivation, spraying, fertilization, irrigation, and harvesting in crop production [8, 27]. Taylor and Fulton [28] reported that several commercially available sensor-based, variable rate systems exist for efficiently managing inputs to maximize yields or returns. They presented a schematic view of a sensor-based, variable rate application system for liquid products as in **Figure 2**. A robot, which apply a microdose of herbicide systematically targeting the weeds that have been detected, is seen in **Figure 3**. It is stated that the total use of herbicides is reduced by as much as 20 times by using this robot because it detects weeds and then targets weeds by moving independently through the field with the help of a camera, GPS sensor, and a solar drive [29]. Remote sensors are usually used to track the soil conditions and plant health. The data obtained from this sensor allows farmers to selectively use (i.e., precisely apply) the exact amount of nutrients, resources, or pesticides necessary for their fields. Grisso et al. [30], who reviewed the variable rate application devices available on the market, discussed the technologies best fit for a cropping system and production management strategy. They presented an "On-the-go" sensor, which measures soil characteristics such as soil moisture content, texture, electrical conductivity, or soil organic matter before planting and adjusting the

*A Review of the Factors Affecting Adoption of Precision Agriculture Applications in Cotton… DOI: http://dx.doi.org/10.5772/intechopen.114113*

#### **Figure 2.**

*Schematic of a sensor-based, variable rate application system for liquid products [28].*

**Figure 3.**

*A robot, which applies a microdose of herbicide by systematically targeting the weeds detected [29].*

seeding rate (plant population) as seen in **Figure 4**. Also, drones and unmanned aerial vehicles systems (UAVs) are used in PA for a variety of tasks such as soil and crop analysis, fertilizer, and pesticide application. Various imaging technologies like hyperspectral, multispectral, and thermal cameras are used in those vehicles,

#### **Figure 4.**

*"On-the-go" sensor (texture, electrical conductivity (EC), or soil organic matter (SOM)) measures soil characteristics before planting and adjusting the seeding rate [30].*

**Figure 5.** *Use of drones and unmanned aerial vehicles systems (UAVs) in precision agriculture [31].*

in order to detect and monitor temporal and site-specific changes in plant health and physiology caused by biotic and abiotic stresses [31]. The use of drones and UAVs in precision agriculture is seen in **Figure 5**.

Gemtos et al. [32] stated that a PA system could be divided into three different phases as the acquisition of weather, soil, and crop data; the processing and analysis of the data; and the implementation and adaptation of cultivation practices.

In PA applications, relative observation data and agronomic models were used in order to implement the applications related to tillage and irrigation scheduling, fertilizer management, weed and pest control, soil and crop growth monitoring, and yield estimation. The main aim of PA applications is to apply the targeted rates of fertilizer, seed, and chemicals for soil, crop, and weather conditions by using site-specific knowledge. PAT enable visualization of spatial and temporal variations between fields or within one location and support spatially varying treatments using variable rate application technologies (VRT) installed on farm agricultural field machinery [8, 16, 17, 33]. Variable rate application (VRA) is one of the most important and recent technologies that have been developed recently to accomplish PA. VRA in PA can
