Sampling design for irrigated vineyards in the riverland of South Australia

Abstract

Two vineyards in the Riverland of South Australia were compared in a geostatistical study. The variables investigated were topsoil depth and root zone readily available water. The first vineyard is developed on highland soils which have a characteristic dune swale formation. The topography of the second is mixed, varying from river flat, through first slope, to highland. Semivariograms from the two vineyards differ substantially. In the first, the correlation between data along the strike of the dunes decreases much less rapidly than across the strike. However, the mixed topography of the second gives semivariograms which are isotropic. The overall level of variability is similar in both cases. Kriged estimates of the values of the variables for irrigation valve areas are sufficiently accurate for irrigation management objectives using the current sampling grids of 75 m × 75 m and 75 m × 50 m for the respective vineyards. A reduction in sample spacing along the strike of the dunes by a factor of 3 would be acceptable for the first vineyard. However, the smaller number of samples in the second vineyard implies that the semivariogram would be inadequately defined if the grid were reduced.