2.4 Data Analyses
To meet the criteria for a dry-down event, the LWD sensor had to
register 100% saturated and decline to the threshold dry level of 15%
without an increase of at least 2% wetness over ten minutes. This
stipulation ensured that an errant water droplet did not discount a
dry-down event but eliminated breaks in precipitation during long storm
events. These inclusion criteria were applied across sites and years,
and a total of at least nine dry-down events that were representative of
each vegetation type were used in the analysis. The data below the
dryness threshold were rescaled as 0% wet.
We also compared the rate of leaf drying between crop sites and forested
conditions. For each 10-minute time point throughout the dry-down event,
the differences in leaf wetness between sites were compared across all
replicate events using a paired t-test (α-value of 95%). Dry-down
curves were best-fit with logarithmic trend lines using the format y =
–m(ln)x + b where m is the slope and b is the y-intercept. We compared
mean least squared differences in dry-down slopes to canopy height using
a back-transformed linearized equation derived from the logarithmic
trends. To allow transformation, we first increased all the data by two
orders of magnitude and converted the dry-down rates to positive values.
Analysis of transformed data reduced variance and kept the resulting
data within the same order of magnitude.
To compare drying times between vegetation types, an analysis of
variance (ANOVA) model was applied for all paired treatment
combinations. Afterwards we applied a Tukey’s post-hoc test to separate
the conditions into like groups by comparing their means. Drying times
were considered different between groups at p < 0.05, and
coefficient of determination (R2) was used to indicate
model fit.
Results