Assertion of Soil Data Consistency by Detecting and Removing Spatial Outliers Using Iterative Techniques for Precision Agriculture

Abstract

In Precision Agriculture, a Decision Support System (DSS) is the ultimate stage that incorporates the basic findings derived from earlier procedures. In many cases, a DSS that is exclusively focused on data-driven methodologies might be highly influenced by data sources if these data resources do not provide any sense of intended conclusion. As a result, on-farm experiment inputs may result in incorrect site-specific crop management in the end. Since data is an essential element of DSS, irregular patterns such as outliers in spatial data that can change the nature of expected outcomes must be avoided during data-driven manipulations. Many of the approaches developed to detect outliers were not designed to deal with masking and swamping effects. With this consideration, the work presented here uses two iterative techniques to locate and remove spatial outliers based on their neighbourhood relationship. As a result of this technique, the masking and swamping effects are reduced. The methods we use are iterative, implying that each iteration discovers and eliminates the expected number of outlying observations. R-Studio is used to demonstrate the use of iterative approaches. The efficacy of both iterative procedures was analysed and compared using one of the current graphical approaches, such as the semivariogram. The research specifically looks at how well these strategies perform on a real-world dataset incorporating spatial observations. The statistical iterative techniques outperformed the graphical approach, according to the findings.

Keywords

• Decision support system (DSS)
• masking and swamping effects
• precision agriculture (PA)
• site-specific crop management (SSCM)
• spatial autocorrelation
• spatial neighborhood