Abstract
Regionalization of an agricultural area by dividing it into different clusters is an important strategy in the precision agriculture scope. Multivariate and spatial data are common in the design of these divisions. This paper sought to characterize regional differences in the area under study through different subsets of variables formed by soil physical-chemical variables and vegetative indices, in an agricultural area for four soybean harvest years in the period from 2013/2014 to 2016/2017. To such end, three subsets were generated comprised by these variables, which presented spatial dependence and were grouped according to their characteristics. By means of decision trees, it was identified which of these variables exerted the most influence on subdivision of the area. The multivariate and non-parametric spatial clustering technique was used to generate the clusters. Finally, by means of maps and boxplots, the spatial relationships between these variables and soybean productivity were evaluated. There was variation across the harvest years in relation to the subset of variables that determined the best design of the different clusters. The regional differences determined by the different variables used in the study showed no relationship with soybean productivity, which presented spatial homogeneity in its data for the harvest years evaluated. This approach is recommended when there is high spatial variability of factors that exert impacts on productivity, advising on using both soil physical-chemical variables and the vegetative indices to explain the causes of soybean productivity spatial variability
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call