df ) for the total NH3
-N loss as a percentage
of applied NH4
-N.
The variability among duplicate measurements
of the hourly rate of NH3
volatilization
was found to be correlated with the average
rate of NH3
loss, a situation also encountered
by Thompson and Meisinger (2002). Linear
regression analysis of the sd of the NH3
volatilization
rate versus the mean rate for each
sampling interval gave a highly significant (p
< 0.01) relationship (r 2
= 0.61) between the
sd and mean. Consequently, rates of NH3
loss
were separated into low-loss, medium-loss,
and high-loss groups to allow pooling of
similar variances and comparison of application
methods. The pooled variances for a
typical 6 hr exposure for the low-loss group
with a mean rate of 45 g NH3
-N ha-1 6 h-1
(0.04 lb NH3
-N ac-1 6 hr-1) produced a sd of
±67 g NH3
-N ha-1 6 h-1 (±0.06 lb NH3
-N
ac-1 6 hr-1), the medium-loss group with a
mean rate of 318 g NH3
-N ha-1 6 h-1 (0.28
lb NH3
-N ac-1 6 hr-1) had a sd of ±253 g
NH3
-N ha-1 6 h-1 (±0.23 lb NH3
-N ac-1 6
hr-1), and the high-loss group with a mean
rate of 1,314 g NH3
-N ha-1 6 h-1 (1.17 lb
NH3
-N ac-1 6 hr-1) yielded a sd of ±462 g
NH3
-N ha-1 6 h-1 (±0.41 lb NH3
-N ac-1 6
hr-1); with each sd having 14 df. These sds
provide a general measure of the uncertainties
in the hourly NH3
-N loss rates
summarized in figures 3 and 4.