Selection of Variables to be Used in Statistical Analysis of Field‐Measured Soil Water Content

Abstract

AbstractInterpretations and inferences drawn from soil water content measurements collected in agronomic field studies are dependent upon the particular method of data analysis. The purpose of this study was to examine alternative approaches for analyzing soil water content data measured by neutron attenuation. Data from tillage studies conducted at two locations on the Atlantic Coastal Plain were used to demonstrate and compare the methods of data analysis. Analysis of variance was performed on the field‐measured data (total soil water content measured by neutron attenuation) and on either six or eight relevant soil water variables derived from them. Derived variables included (i) the net change in total water content between successive measurement dates; (ii) a transformed soil water content; and (iii) six variables derived using bulk density and 15 bar soil water contents to provide estimates of plant‐available water.The total soil water content variable has a coefficient of variation (C.V.) of about 12%, and the cost of collecting data and analyzing it is relatively low. Differences in total water content between successive dates have C.V.s generally about 100%. Requiring the measurement of the 15 bar percentage and bulk density values, which allows plant‐available water to be calculated, increases the costs of data collection. Incorporation of this information into the water variables results in data which have a C.V. of about 30%. In field experiments, the investigator should select the appropriate soil water variable only after examining the relative costs, expected C.V.s, and the anticipated use of the data.