1. Introduction
A physical environment has a dynamic behavior, as it is an open system and involves variable driving forces that interact with each other in a spatial and temporal scale [
1]. An anthropogenic environment is another dynamic system, which intervenes to the functions of the physical environment and highly influences its elements. A typical example of this co-existence is the field of agriculture, as farmers are called to work in the physical environment and combine agronomic management practices [
2]. More specifically, in crop cultivation, farmers ought to manage soil, water, and crop resources together with applying inputs and practices, in order to yield maximum quality products without burdening the environment. Soil degradation under climate change, such as evolvements in arid or semi-arid environments (e.g., in south–eastern Europe), imperatively demand the development of methodologies and planning for reversing current situation [
3,
4].
When referring to space and time, the ways of practicing crop management vary, with the most common being conventional and site-specific. Conventional farming is mostly characterized by uniform management strategies [
5]. This means that every decision taken during a growing period is homogenously applied across a field. For example, it is common practice for farmers to apply the same fertilizer rates across their fields. On the other hand, site-specific crop management treats an agricultural area as a non-homogenous field in terms of soil, crop, and yield characteristics [
6]. Site-specific crop management is a well-known set of agronomic practices, rationalizing use of inputs mainly motivated by economic and environmental purposes [
5,
7]. Site-specific management does not ignore the expert knowledge of the farmer, but incorporates it in the decision process, taking into consideration the fact that soil and crop characteristics differentiate over space and time.
As stated in [
8], spatial variability is a natural and inevitable characteristic of all soil bodies. Soil physicochemical properties are not spatially and temporarily stable; hence, anyone who deals with soil resources and relies on their productivity ought to take into account this inconsistency. Soil characteristics may vary even inside the boundaries of a single field. However, in order to adjust farming management options to soil variability, there should be a robust and trustworthy methodology for measuring or even estimating this variability. In [
9], the authors developed and tested a soil strength sensor in field conditions, capable of characterizing within-field soil variability. Further, soil electrical conductivity is thought to be suitable for detecting spatial patterns in soil characteristics, as it correlates with soil texture and moisture [
10]. In [
11], the authors used soil proximal sensors, data fusion, and mining methodologies to predict soil organic matter, soil acidity, lime buffer capacity, calcium, magnesium, and aluminum, towards understanding of soil heterogeneity. In [
12], the authors developed an innovative and multifunctional analysis system, which provides quantitative and precise information about the properties of soil profiles, offering a promising means of measuring and understanding soil variability. Soil threats (e.g., soil vulnerability to erosion) have also been modeled through sensing systems and artificial intelligence (AI) methodologies [
13].
Additionally, as crops are cultivated in such variable soil environments, it is obvious that their nutritious and yield characteristics are also space and time dependent. In the work of [
14], the spatial variability in grape yield through field zoning was analyzed, and soil and crop factors that affect this variability were determined. The research team in [
15] correlated key soil properties, such as soil organic carbon, soil acidity, and clay content with micronutrients and trace metals, proving their spatial correlation. The dependence of crop yield variability on soil factors was investigated by [
16] in a four-year experiment concerning cereals. They proved that the fluctuation of soil electrical conductivity might represent soil variability and lead to yield variations, denoting that site-specific management should be annually evaluated, as variability is not constant over the years. In [
17], the authors managed to produce digital surface maps with the use of an unmanned aerial system (UAS) representing accurately the spatial variability of maize plant heights throughout a growing season.
Under the framework of circular economy (CE), crop production ought to deliver strategies that promote food quality, sustainability, pollution mitigation, waste reuse, etc. CE provides a broad spectrum of actions for achieving better and long-term living standards [
18,
19]. The CE concept can be applied at multiple levels of industrial and agricultural systems [
20,
21]. The sustainability of municipal solid waste management practices and procedures was assessed through a model-based analysis in [
22], revealing various grades of convergence towards CE. Furthermore, food waste can be used for energy recovery, and under special treatment, it can lead to diminished waste load [
23]. Regarding waste management, a comprehensive framework for conceptualizing and implementing a strategy is analyzed in [
24], while the evaluation of environmental performance with the use of key indicators is realized in [
25], which may also provide insights and guide strategic planning in crop management. Soil-aquifer pollution and agricultural waste strategies receive great attention in literature [
26,
27,
28] while site-specific crop management may positively contribute to applying those managerial plans in an effective and sustainable way [
29].
Remote sensing technology is a very common means, not only for monitoring crop variability [
30,
31,
32], but also for environmental applications in general [
33]. The evolution of small, unmanned aerial vehicles (UAVs) integrated with a variety of payload sensing systems, which are available in low-cost and mass production hardware solutions, has led to their wide use in agricultural research experimentation [
34]. Indeed, many aerial datasets have been acquired via UAVs, and analyzed, correlating spectral signatures with parameters of field conditions [
35]. The spatial nature of these datasets, along with the point reference of soil and crop field measurements, make the use of geospatial methodologies apparent, as far as their analysis is concerned [
36]. Geographic information systems (GIS) offer an integrated suite of software and hardware tools for managing spatiotemporal data and mapping in-field variability [
37,
38]. Many research works have combined the aforementioned methods, mostly for evaluating soil conditions [
39,
40,
41], while monitoring in situ crop nutrient variations is not so commonly met in scientific literature. Several studies have also used geospatial analysis for separating a field into management zones, concluding that this practice can well manage sub-field soil and crop variability [
42,
43,
44].
The main purpose of conducting the present study was to investigate possible interrelations between spectral and field data under the concept of site-specific crop management. Towards this objective, a commercial low-cost UAV was used to capture orthophotographs in near-infrared wavelength and classic field sampling and analyses procedures were incorporated to determine soil and crop parameters in cotton crops. Consequently, statistical and geostatistical methods combined with the fuzzy algorithm application, undertook the correlation and the assessment of the outputs. A further scope was to check the feasibility of using a UAV in order to delineate in-field management zones based on soil, crop, and reflectance data.
4. Conclusions
In the context of this study, UAS data were analyzed with soil and crop parameters in two cotton fields during a growing period. All data were analyzed using geostatistical and geospatial methodologies coupled with PCA and Fuzzy c-means clustering under GIS environment. A primary conclusion, of practical importance is that the use of UAS in crop production offers a quick and reliable way to monitor soil and plant capital. As many UAS manufacturers throw their products into the market, raising competition, and diminishing costs, many farmers are able to engage UAS or similar services to their production procedure. Another finding is that fieldwork, in the form of sampling and further analyzing soil and tissue samples, are of crucial importance for evaluating and confirming remotely sensed data. This work demonstrated that reflectance data were correlated with organic matter, carbonate, and clay content, while those data can be directly used for in-field zone delineation. This time efficient form of field monitoring can lead to targeted sampling schemes, avoiding unnecessary work and costs. On the contrary, crop nutrient characteristics were insignificantly combined with aerial data, which may have been the result of uniformity of the measured parameters across the two fields. The resulting zones, as mapped in this case, offer little in fertilizing management guidance. Further work should evaluate present results by incorporating more fields, crops, and growing seasons. Research efforts should also focus on translating the outcomes of soil and crop monitoring through expert decision-making tools and on efficiently applying management plans through variable rate technologies. The integration of new technologies and established methodologies in primary production, such as those demonstrated, provide notable means for applying site-specific crop management in broader adoption levels and constitute a crucial motive for visualizing, designing, implementing, and assessing environmental strategic plans towards a circular economy.