SPATIAL VARIABILITY AND SAMPLING STRATEGIES FOR NO3-N, P, AND K DETERMINATIONS FOR SITE-SPECIFIC FARMING

Abstract

NO3-N, P, and K nutrients data were analyzed to examine the extent of spatial variations in theirconcentration levels and to develop sampling strategies to obtain their representative mean concentrations and correlatedcharacteristics in spatial domain. The study was conducted at two farms located within the Coastal Plains of Virginia.The nutrients concentration values varied with coefficients of variation ranging from 13% to 40% at the two sites. Spatialvariations along and across the rows at study site were the same at Brandon farm, but were different at the Renwoodfarm. The sample size required to determine the representative mean values of NO3-N and K nutrients could be decreasedby about 40% to 60% if the associated error in the mean values was increased from 5% to 10% level. The concentrationvalues of three nutrients exhibited spatially correlated dependence on their adjacent measurements. The correlationstructures of nutrients could best be described by the exponential forms of the semi-variogram model for each site, exceptfor K at Brandon site for which a spherical model was found to be more appropriate. Based on the spatial variability andcorrelation length of nutrients, sampling grids of 32 32 m size for NO3-N, P, and K determinations at the Brandon site,and of 18 32 m for NO3-N, 29 59 m for P, and 23 44 m for K determinations at the Renwood site are recommended.However, due to cost and practical considerations, sampling grid of 18 32 m size for the three nutrients is suggested forthe Renwood site. A reduction of 12% in P fertilizer at Brandon and of 25% in K fertilizer at Renwood site could beachieved if fertilizer is applied at a variable rate based on the spatial variability of the nutrient levels of the soil.However, the requirement for K fertilizer for Brandon and P fertilizer for Renwood site would be approximately 6%higher compared to those for the conventional strategy. The information presented here could be used as a guideline forrefining the existing sampling practices for implementing the variable-rate fertilizer application technology. However, itseconomic feasibility is yet to be studied.