Assessment of the optimal number of observations in the study of vineyard soil (Rigosol)

A study of soil pH on the experimental field resulted in a high variability of pH on a very small scale. This kind of heterogenity in soil pH have effects on growth of two grapevine varieties on rootstock Kober 5BB: Riesling and Pinot Noir A number of 104 soil samples were taken from an area of 1.43 ha from two depths. A goal of this experiment was to find the optimum number of samples for pH studies, and to implement the obtained results in further investigation on experimental fields. Therefore, in this paper we compared diferent deterministic interpolation techniques: inverse distance weight, splines and local polynomial interpolation, on the results of soil pH. Root mean square error (RMSE) statistitics obtained after cross validation procedure was used for the choice of appropriate exponent value for IDW, spline and local interpolation. The obtained interpolation parameters were used for mapping the field and the most accurate technique was IDW, which was further used in creation of pH maps with lower number of samples: 54, 34, 29, 24, 19 and only 14 pH samples. Maps were classified and compared by means of percentage difference in area among classes of pH in respect to classes obtained after maximum sampling. The results indicated that the criteria of 15% of change in pH area over classes could be satisfied with only on third of the samples. An obtained results will be used for further sampling of the whole experimental area.