ERROR ANALYSIS IN ESTIMATING SOIL WATER CONTENT FROM NEUTRON PROBE MEASUREMENTS

Publisher Website
Google Scholar
Abstract
Variance analysis is presented for identifying the different components (instrument, calibration, and location) of the errors involved in estimating the spatial mean water content values from neutron probe readings. Soil moisture was measured by 2 neutron probes at 17 locations regularly spaced on the field. For each probe, unique calibration curve was used for the whole field with both biased and unbiased treatments of the linear correlations. All the measurements at a specified depth were normally distributed regardless of their location, and no spatial structure has been shown for the sampling scheme. In situations of site homogeneity or a large number of sampling points, the calibration component is the greatest contribution to the total variance. It is reduced by considering the unbiased statistical treatment of the neutron probe calibration curve. For higher levels of field heterogeneity, the location component becomes the major contribution to the variance of mean values. It can be decreased by increasing the number of sampling points. Finally, the suitable number of sampling points necessary to provide mean water content values with a prescribed precision is given for the experimental site by applying the central limit theorem.