| Abstrakt: |
Soil samples were collected from nine fields during 2018-2020 to evaluate different soil sampling methodologies, including two of the most popular interpolation methods. The interpolation methods used in this study were inverse distance weighting (ID) and Kriging (Kr). The ID method assumes that soil samples close to one another are more alike than those farther apart. In contrast, the Kr method considers the distance and how much variability exists among known soil sampling locations. Soil samples were collected in a 1-acre grid fashion in nine fields across Arkansas, representing different soil series and crop rotation practices. It appears that the choice of interpolation method may not only affect the total amount of nutrient recommended but also how the nutrient is distributed across a field. Empirical semivariograms were fit to both raw and log-transformed data using ArcGIS Geostatistical Analyst (ESRI, Redlands, Calif.), with Stable, Gaussian, and spherical (only for non-transformed data). The fitted model's selection was mainly based on which model had a resulting root mean squared standardized errors (RMSE) closest to 1. Semivariograms were used to determine the range, which is the distance that assures independent readings. Based on the spatial variability of the nutrients in the fields sampled as a part of this study, the estimated ranges for phosphorus (P) varied between 276 and 801 feet, corresponding to sampling grids of about 1 to 4 acres. In comparison, the calculated range values for potassium (K) varied between 401 and 1495 feet, which corresponds to sampling grids of about 2 to 8 acres. Apparent electrical conductivity (ECa) values were obtained to test the relationship between nutrient concentrations and ECa values. Results showed that, for the fields sampled as part of this study, the history of variable-rate fertilization appears to lessen the effect of soil type on nutrient concentration. [ABSTRACT FROM AUTHOR] |
