On the issue of semivariograms constructing automation for precision agriculture problems

Abstract

Most precision agriculture (PA) problems are based on assessing the variability of agroecological parameters (differentiated fertilizer application, allocation of heterogeneous zones of the agricultural field, etc.). In this connection, the study of the spatial structure of such data using geostatistical methods seems to be a relevant and promising direction. To validate the semivariogram model, determine its parameters, analyze anisotropy, and carry out other stages of the experiment, a large number of numerical calculations are required. Manually all these calculations are extremely difficult to perform, therefore, automation of these processes is necessary. Most existing programs exclude certain actions that may be important in solving PA problems (there is no possibility of studying anisotropy or spatial trend, the number of theoretical variogram models, etc., is limited), and the use of programming languages (R, Python, etc.) requires deep expertise. Therefore, there was a need to automate the solution of a certain range of problems by geostatistical methods. For the implementation of the module under consideration, it seems optimal to use the R programming language, which has a number of significant advantages: open source code and free access, a large number of supported and regularly updated packages, wide graphical capabilities, cross-platform, etc. General suggestions for automating the construction of semivariograms are presented and further use in solving a certain range of PA tasks.