Sediment Yield Methods
NTU- and Q-based (loge transformed) modeled SSCs
(converted from mg L-1 to mg m-3)
were multiplied by discharge separately to give sediment yield (mg
s-1) and converted to integrate mass over longer
periods (Mg yr-1). Confidence and prediction intervals
for modeled SSCs were calculated in R, then converted to lower and upper
yields using the same method. NTU- and Q-based half-hourly or hourly
sediment yields (depending on the interval of model parameters) were
bridged by taking the period of available NTU-based yields and
subsequently filling gaps where NTU-based yield values were not
available with Q-based yield values, to give continuous records of
sediment yield for both Carnivore and Chamberlin Creeks for the majority
of the 2015 and 2016 open-channel seasons. For early- and late-season,
when modeled SSCs could not be predicted with either NTU- or Q-based
models due to lack of data, average sediment yield below the
corresponding estimated low-flow discharge estimated from photographs
and field notes was used (< 10 m3s-1 and < 5 m3s-1 in Carnivore Creek; < 0.25
m3 s-1 in Chamberlin Creek).
Subsequently, sediment yields for the entire open-channel seasons in
both creeks could be estimated. We used May 18 to September 17 as the
open-channel period, based on data availability and field photographs.
During mid-May when flows were ice affected, field observations and
photographs indicate that our dataset may exclude up to one week of
low-flow water and sediment discharge. We are confident that no
significant late-season events were missed based on photographic
evidence and regressions of Q from our study with the nearby Hula Hula
River discharge (U.S. Geological Survey gauge 15980000), with the latter
explaining about half the late season Q variability.
Results