This research studies the effect of stratifying soil samples to try and find a suitable depth to establish a geospatial relationship for a practical soil sampling grid in New Zealand hill country. Cores were collected from 200 predetermined sites in grids at two trial sites at "Patitapu" hill country farm in the Wairarapa, New Zealand. Trial 1 was a 200 m × 100 m grid located in a gently undulating paddock. Trial 2 was a 220 m × 80 m grid located on a moderately sloped paddock. Each grid had cores taken at intervals of 5 m, 10 m, or 20 m. Core sites were mapped out prior to going into the field; these points were found using a Leica Geo Systems GS15 (real time kinematic GPS) and marked with pigtail pegs and spray-paint on the ground. Cores were taken using a 50 mm-diameter soil core sampler. Cores were cut into three sections according to depth: A-0-30 mm, B-30-75 mm, and C-75-150 mm. Olsen P lab results were obtained for half of the total 1400 samples due to financial constraints. The results indicate that there was a significant decrease in variability from Section A to Section B for both trials. Section B and C for Trial 1 had similar variability, whereas there was another significant drop in variability from Section B to C in Trial 2. Measuring samples below the top 3 cm appeared to effectively reduce noise when sampled from 3 to 15 cm. However, measuring from 7.5 cm to 15 cm on the slope in Trial 2 reduced variability so much that all results were almost identical, which may mean that there is no measurable representation of plant available P. The reduction in noise by removing the top 3 cm of soil samples is significant for improving current soil nutrient testing methods by allowing better geospatial predictions for whole paddock soil nutrient variability mapping.

This work examines two large data sets to demonstrate that hyperspectral proximal devices may be able to measure soil nutrient. One data set has 3189 soil samples from four hill country pastoral farms and the second data set has 883 soil samples taken from a stratified nested grid survey. These were regressed with spectra from a proximal hyperspectral device measured on the same samples. This aim was to obtain wavelengths, which may be proxy indicators for measurements of soil nutrients. Olsen P and pH were regressed with 2150 wave bands between 350 nm and 2500 nm to find wavebands, which were significant indicators. The 100 most significant wavebands for each proxy were used to regress both data sets. The regression equations from the smaller data set were used to predict the values of pH and Olsen P to validate the larger data set. The predictions from the equations from the smaller data set were as good as the regression analyses from the large data set when applied to it. This may mean that, in the future, hyperspectral analysis may be a proxy to soil chemical analysis; or increase the intensity of soil testing by finding markers of fertility cheaply in the field.