Spatial Variability of Soil Nitrogen and Considerations for Soil Sampling for Precision Agriculture.

Soil sampling is critical to N management in precision agriculture. Here we address some practical questions: 1) How many cores (subsamples) are needed to obtain a reliable composite sample given a fixed sample support? 2) How large of an area (sample support size) can a composite sample represent given a fixed number of cores? 3) Can we obtain a reliable estimate of soil properties at 0-30 or 0-60 cm depth with samples collected at 0-15 cm depth in order to reduce the sampling cost? Two systematic sampling schemes in a farm field in Minnesota were designed: 1) transect sampling at 0-30 cm, in which two transects were targeted and along each transect 200 cores (1.9 cm diameter) were collected at an interval of 15 cm (i.e. the transect length is 30m).  This high-density sampling reveals the fine scale variability at core-size support to address the first two questions. 2) composite sampling, in which 10 cores were randomly collected in 180 3x3 m areas at depths of 0-15, 15-30, and 30-60 cm and 10 cores in each layer were combined to make a composite sample. All samples were measured for total nitrogen, total inorganic nitrogen, and total organic carbon. Semivariogram analysis was used to characterize the spatial variability of soils and spatial linear mixed model to predict soil properties at 0-30 and 0-60 cm depths with 0-15 cm observations. Preliminary results show that samples needed to obtain estimates with 10% error at 5% significance level varied across soil properties but the 0-15 cm depth could be a good predictor for 0-30 cm depth; however, the 0-15 cm depth has limited predictability for the 0-60 cm depth. This research will not only advance our understanding of multiple-scale soil spatial variability but also provide useful guidance for soil sampling to support precision agriculture.