Electronic estimation of hay moisture content: precision and accuracy

Abstract

Listen

Translate article

Four commercial meters were used to measure hay moisture content (MC) in the range of 10 to 40% (on a wet basis). Three meters (#1, 2 and 4) were based on electrical conductance; the other meter (#3) was based on capacitance. The standard MC was obtained by oven drying at 103C during 24 h, according to ASABE Standard 358.2. Results showed that the standard deviation (SD) of MC increased with MC. When SD was divided by MC, the resulting coefficient of variation (CV) was not significantly affected by MC. For the standard MC of hay obtained by oven drying, the CV averaged 6.0%. Two meters (#1 and 3) had a CV of 8.4% while the two other meters had a CV of 10.7%. The approximate confidence interval (CI) for the standard MC was 0.90% at 15% MC and 1.50% at 25% MC (based on 5 samples). Similarly, the two meters with the lowest CV had an approximate CImeter of 2.24% at 25% MC (based on 5 samples) and the two other meters had a CImeter of 2.84% at 25% MC. Linear regression was used to correlate the meter readings with the standard MC. The capacitance type meter (#3) had the highest coefficient of determination (R = 0.954 at low hay density); however, its error of estimate varied by almost 8% between 15 and 25% MC, thereby requiring important corrections of reading. One of the electrical conductance meters (#1) had a small error of estimate change (0.4% over the range of 15 to 25% MC), thereby requiring only small corrections of reading to estimate the true MC. When combining the error resulting from meter variability and the error of estimate of true MC, a global CI was estimated which ranged from 1.16% to 2.14% depending of MC measured and the number of measurements made (5 or 10).